HHS Public Access Author manuscript Author Manuscript
Neuron. Author manuscript; available in PMC 2017 September 22. Published in final edited form as: Neuron. 2017 August 16; 95(4): 928–943.e3. doi:10.1016/j.neuron.2017.07.028.
LTP at Hilar Mossy Cell-Dentate Granule Cell Synapses Modulates Dentate Gyrus Output by Increasing Excitation/ Inhibition Balance Yuki Hashimotodani1,2,5, Kaoutsar Nasrallah1,5, Kyle R. Jensen1, Andrés E. Chávez1,3, Daniel Carrera1,4, and Pablo E. Castillo1,6,*
Author Manuscript
1Dominick
P. Purpura Department of Neuroscience, Albert Einstein College of Medicine, Bronx, NY 10461, USA
SUMMARY
Author Manuscript
Excitatory hilar mossy cells (MCs) in the dentate gyrus receive inputs from dentate granule cells (GCs) and project back to GCs locally, contralaterally, and along the longitudinal axis of the hippocampus, thereby establishing an associative positive-feedback loop and connecting functionally diverse hippocampal areas. MCs also synapse with GABAergic interneurons that mediate feed-forward inhibition onto GCs. Surprisingly, although these circuits have been implicated in both memory formation (e.g., pattern separation) and temporal lobe epilepsy, little is known about activity-dependent plasticity of their synaptic connections. Here, we report that MCGC synapses undergo a presynaptic, NMDA-receptor-independent form of long-term potentiation (LTP) that requires postsynaptic brain-derived neurotrophic factor (BDNF)/TrkB and presynaptic cyclic AMP (cAMP)/PKA signaling. This LTP is input specific and selectively expressed at MCGC synapses, but not at the disynaptic inhibitory loop. By increasing the excitation/inhibition balance, MC-GC LTP enhances GC output at the associative MC-GC recurrent circuit and may contribute to dentate-dependent forms of learning and epilepsy.
INTRODUCTION The dentate gyrus, the principal input region of the hippocampus, plays a key role in memory formation by transforming patterns of cortical inputs into new patterns of output to the CA3 area (Kesner and Rolls, 2015; Knierim and Neunuebel, 2016). Although the cellular
Author Manuscript
*
Correspondence:
[email protected]. 2Present address: Department of Neurophysiology, Graduate School of Medicine, The University of Tokyo, Tokyo 113-0033, Japan 3Present address: Centro Interdisciplinario de Neurociencia de Valparaíso, Facultad de Ciencias, Universidad de Valparaíso, Valparaíso 2340000, Chile 4Present address: National Institute of Neurological Disorders and Stroke, National Institutes of Health, Bethesda, MD 20892, USA 5These authors contributed equally 6Lead Contact SUPPLEMENTAL INFORMATION Supplemental Information includes six figures and one table and can be found with this article online at http://dx.doi.org/10.1016/ j.neuron.2017.07.028. AUTHOR CONTRIBUTIONS Y.H., K.N., K.R.J., and A.E.C. performed research and analyzed the data. D.C. performed initial BDNF puff experiments. Y.H., K.N., K.R.J., A.E.C., and P.E.C. designed the experiments. Y.H, K.N. K.R.J., and P.E.C. wrote the manuscript. All authors read and edited the manuscript.
Hashimotodani et al.
Page 2
Author Manuscript
and synaptic basis of this transformation remains unclear, the two glutamatergic cell types in the dentate gyrus, granule cells (GCs) and hilar mossy cells (MCs), likely play a major role. GCs receive excitatory inputs from the entorhinal cortex via the perforant path (PP) and send excitatory output to CA3 pyramidal neurons via the mossy fibers (Amaral et al., 2007). MCs mediate an intrinsic (or associative) excitatory loop, receiving powerful input from a relatively small number of GCs and providing highly distributed excitatory output to a large number of GCs (Amaral et al., 2007; Buckmaster and Schwartzkroin, 1994; Buckmaster et al., 1996; Scharfman and Myers, 2013). In addition to the recurrent circuit, MCs also contact GABAergic interneurons, which mediate feed-forward inhibition onto GCs (Larimer and Strowbridge, 2008; Scharfman, 1995). Although MCs were first identified over a century ago (Lorente De Nó, 1934; Ramón y Cajal, 1911), there are still significant gaps in our knowledge about their function (Scharfman, 2016), and little is known about activitydependent plasticity of their synaptic outputs.
Author Manuscript Author Manuscript
MCs project their associational and commissural axons to the ipsi- and contralateral inner molecular layer (IML) of the dentate gyrus, where they synapse onto proximal dendrites of GCs (Scharfman, 2016; Scharfman and Myers, 2013). Because of their proximity to the GC soma, MC-GC synapses are in an ideal position to influence the activity of GCs. Moreover, MCs not only contact GCs locally (same lamella) but also project widely along the longitudinal axis of the hippocampus, both septally and temporally from the point of origin (Amaral et al., 2007; Buckmaster et al., 1996). It has been estimated that a single MC may innervate as much as 75% of the septotemporal axis (Amaral and Witter, 1989) and establish ~35,000 synapses in the IML onto putative GC dendrites (Buckmaster et al., 1996). The hippocampus is functionally heterogeneous along this axis; the dorsal/septal hippocampus is primarily involved in spatial memory, while the ventral/temporal hippocampus is associated with emotional memory (Fanselow and Dong, 2010; Strange et al., 2014). Thus, the translamellar projection of MCs could modulate GC activity throughout the hippocampus, thereby linking functionally diverse areas (Scharfman and Myers, 2013). Based on the wide distribution of their axons along the septotemporal axis and remarkable divergence onto GCs, MCs likely play a major role in dentate gyrus information transfer. Furthermore, activity-dependent plasticity of MC-GC transmission is expected to have a significant impact on dentate-gyrus-dependent learning.
Author Manuscript
MCs, via their communication with GCs, have been implicated in various forms of learning and memory, including associative memory (Buckmaster and Schwartzkroin, 1994), pattern separation (Myers and Scharfman, 2009), and recall of memory sequences (Lisman, 1999; Lisman et al., 2005). Additionally, MCs could contribute to temporal lobe epilepsy (Scharfman and Myers, 2013), either by becoming overactive and driving GC firing (Ratzliff et al., 2002), or by dying and reducing the magnitude of feed-forward inhibition (Sloviter, 1991). While long-term synaptic plasticity has been thoroughly characterized at almost every connection of the classical “trisynaptic” excitatory circuit of the hippocampus (Bliss et al., 2007), only a handful of in vivo studies, with mixed results, have explored the occurrence of long-term potentiation (LTP) at the MC-GC synapse (Alvarez-Salvado et al., 2013; Bekenstein and Lothman, 1991; Hetherington et al., 1994; Kleschevnikov and Routtenberg, 2003; Steward et al., 1990) and, to our knowledge, no in vitro study has characterized any form of long-term plasticity at these synapses. Neuron. Author manuscript; available in PMC 2017 September 22.
Hashimotodani et al.
Page 3
Author Manuscript
In this study, we report that repetitive activation of MC axons induces a robust form of LTP at MC-GC synapses whose induction, unexpectedly, is NMDAR independent. Remarkably, MC-GC LTP requires postsynaptic BDNF/TrkB and presynaptic cyclic AMP (cAMP)/PKA signaling. This form of LTP was selectively induced at the monosynaptic excitatory MC-GC pathway, but not at the feed-forward inhibitory pathway. By changing the excitation/ inhibition (E/I) balance and modulating GC output, MC-GC LTP might contribute significantly to memory formation and also participate in temporal lobe epilepsy.
RESULTS MC-GC Synapses Express Robust NMDA-Receptor-Independent LTP
Author Manuscript Author Manuscript
To investigate whether MC-GC synapses exhibit activity-dependent long-term plasticity, we first performed whole-cell patch-clamp recordings from GCs in acute rat hippocampal slices (holding potential [Vh] = −60 mV, Cs+-based internal solution). MC-induced excitatory postsynaptic currents (MC-GC EPSCs) were elicited by extracellular stimulation in the IML of the dentate gyrus in the presence of 100 μM picrotoxin to block fast inhibitory synaptic transmission (Figure 1A) (Chiu and Castillo, 2008). Repetitive activation of MC axons with brief bursts (5 pulses, 100 Hz, repeated 50 times every 0.5 s) triggered robust LTP of MCGC EPSCs (Figure 1B; 214% ± 16% of baseline, n = 18, p < 0.001, paired t test). To confirm the identity of the evoked EPSCs, we applied the mGluR2/3 agonist DCG-IV (1 μM), which selectively reduces transmission at neighboring medial perforant path to GC (MPP-GC) synapses (Macek et al., 1996), but not MC-GC synapses (Chiu and Castillo, 2008). The magnitude of MC-GC LTP showed strong dependence on the frequency of stimulation and number of bursts (Figures 1C and 1D). Importantly, similar patterns of MC activity that trigger LTP of MC-GC transmission have been recently reported in behaving mice and rats (Danielson et al., 2017; GoodSmith et al., 2017; Senzai and Buzsáki, 2017). We also found robust MC-GC LTP using a more physiological K+-based internal solution (Figure S1A; K+-based internal: 186% ± 18% of baseline, n = 7, p < 0.01), performing noninvasive extracellular field excitatory postsynaptic potential recordings (fEPSPs) (Figure S1B; fEPSPs 164% ± 10% of baseline, n = 6, p < 0.05, Wilcoxon signed rank test), in mouse hippocampal slices (Figure S1C; 186% ± 28% of baseline, n = 4, p < 0.05, paired t test), in the absence of picrotoxin while voltage clamping GCs at ECl− (Figure S1D; 183% ± 19% of baseline, n = 7, p < 0.001, paired t test), and in the presence of the GABAB receptor antagonist CGP55845 (3 μM) (Figure S1E; 200% ± 27% of baseline, n = 5, p < 0.05, paired t test).
Author Manuscript
Classical Hebbian LTP requires NMDA receptor (NMDAR) activation (Bliss et al., 2007). To test whether MC-GC LTP in acute hippocampal slices also relies on NMDAR activation, we delivered the induction protocol in the presence of the NMDAR antagonist D-APV (50 μM). To our surprise, MC-GC LTP was intact (Figure 1E; 193% ± 21% of baseline, n = 12, p < 0.001, paired t test). Moreover, a pairing-protocol commonly used to trigger Hebbian LTP (i.e., presynaptic activity [200 pulses, 2 Hz] paired with postsynaptic depolarization to 0 mV) also induced MC-GC LTP, and this plasticity was equally induced in the presence of 50 μM D-APV (Figure 1F; control 180% ± 12% of baseline, n = 12, p < 0.001, paired t test; DAPV: 186% ± 28% of baseline, n = 7, p < 0.001, paired t test; control versus D-APV: p >
Neuron. Author manuscript; available in PMC 2017 September 22.
Hashimotodani et al.
Page 4
Author Manuscript
0.5, unpaired t test) or in the absence of postsynaptic depolarization (Vh = −60 mV) (Figure 1F; presynaptic [pre] only: 214% ± 21% of baseline, n = 8, p < 0.01, Wilcoxon signed rank test; control versus pre only, p > 0.17, Mann-Whitney U test). In contrast, and as expected for a Hebbian, NMDAR-dependent form of plasticity, D-APV blocked a pairing protocolinduced LTP at MPP-GC synapses (Figure 1G; control: 167% ± 6% of baseline, n = 6, p < 0.001, paired t test; D-APV: 99% ± 9% of baseline, n = 6, paired t test, p > 0.2; control versus D-APV: p < 0.001, unpaired t test), and this LTP was not observed in the absence of postsynaptic depolarization (pre only: 100% ± 3% of baseline: n = 5, p > 0.8, paired t test). Altogether, these results indicate that unlike MPP inputs, MC inputs onto GCs can express an NMDAR-independent form of LTP.
Author Manuscript
We next examined whether presynaptic burst activity alone, which triggers robust MC-GC LTP (as shown in Figures 1A and 1B), could also induce LTP at MPP-GC synapses. Unlike MC-GC synapses, this induction protocol did not trigger LTP of MPP transmission (Figure 1H; 92.4% ± 6.5% of baseline, n = 11, p > 0.2, paired t test). This observation, together with the contrasting NMDAR requirement (Figures 1E–1G), reveals fundamentally different induction rules at MC and MPP synapses. To examine whether MC-GC LTP is input specific, we placed two stimulating pipettes in the IML (~100 μm from the recorded GC) (Figure 1I, left) in order to activate two MC inputs, and we delivered the LTP induction protocol to only one input, while the naive input served as control. We found that only the tetanized input showed LTP (Figure 1I; tetanized: 205% ± 15% of baseline, n = 5, p < 0.005, paired t test; naive: 101% ± 7% of baseline, n = 5, p > 0.4, Wilcoxon signed rank test; tetanized versus control: p < 0.05, Mann-Whitney U test), suggesting that signaling involved in MC-GC LTP did not spread to nearby naive MC-GC synapses. Thus, MC-GC LTP is an input-specific phenomenon.
Author Manuscript
MC-GC LTP Is Expressed Presynaptically
Author Manuscript
MC-GC LTP was accompanied by significant decreases in paired-pulse ratio (PPR; baseline: 1.36 ± 0.05; LTP: 1.07 ± 0.05, n = 18, p < 0.001; paired t test) and coefficient of variation (CV; baseline: 0.33 ± 0.02; LTP: 0.22 ± 0.02, n = 18: p < 0.001; paired t test) (Figure 2A), suggesting a presynaptic form of LTP expression (Castillo, 2012). If MC-GC LTP is due to a long-lasting increase in glutamate release, both AMPAR- and NMDAR-mediated components of synaptic transmission are expected to show a similar degree of potentiation. To test this possibility, we simultaneously assessed these components by monitoring compound AMPAR/NMDAR EPSCs while voltage clamping GCs at −40 mV (Figure 2B) and measuring the AMPAR and NMDAR components at the peak and 40 ms post-stimulus, respectively. Under these recording conditions, both components were equally potentiated (Figure 2B; n = 9, p > 0.2, paired t test), suggesting an increase in glutamate release. Consistent with a presynaptic mechanism of expression, we found that MC-GC LTP elicited by minimal stimulation in IML (Figure 2C) was associated with a significant decrease in failure rate (Figures 2C and 2D; baseline: 53.9% ± 3.7%; LTP: 19.3% ± 5.9%, n = 8, p < 0.001; paired t test), an increase in efficacy (i.e., mean EPSC amplitude including failures) (Figures 2C and 2D; baseline: −8.4 ± 2.0 pA; LTP: −28.7 ± 9.5 pA, n = 8, p < 0.01; Wilcoxon signed rank test), and an increase in potency (i.e., mean EPSC amplitude excluding failures) (Figures 2C and 2D; baseline: −16.2 ± 3.7 pA; LTP: −32.0 ± 10.1 pA, n
Neuron. Author manuscript; available in PMC 2017 September 22.
Hashimotodani et al.
Page 5
Author Manuscript Author Manuscript
= 8, p < 0.01; Wilcoxon signed rank test). The increase in potency is likely due to the multiple contacts that a MC axon establishes with a GC (Buckmaster et al., 1996), which allows for coincident release from multiple release sites. Importantly, the robust MC-GC LTP observed by minimal stimulation (292.1% ± 44.3% of baseline, n = 8, p < 0.01, paired t test) suggests that LTP induction does not necessarily require bulk, synchronous activation of MC axons but can occur as a result of more physiological activity. To discard the potential contribution of a postsynaptic mechanism of expression, we included botulinum toxin-B (BoTx) in the intracellular recording solution, a manipulation that blocks postsynaptically expressed LTP by preventing the SNARE-complex-dependent delivery of new receptors to the synapse (Lledo et al., 1998). BoTx was used at 5 and 500 nM, and data were merged, as no obvious difference was observed between these two concentrations. We found normal MC-GC LTP in BoTx-loaded GCs (Figure 2E; control: 196% ± 18% of baseline, n = 7; BoTx: 180% ± 8% of baseline, n = 9; control versus BoTx: p > 0.8, unpaired t test). BoTx activity was confirmed in interleaved slices by the reduction in AMPAR-mediated transmission in BoTx-loaded CA1 pyramidal cells (Figure 2F; control: 107% ± 8% of baseline, n = 12; BoTx: 57% ± 4% of baseline, n = 11; control versus BoTx: p < 0.001, unpaired t test) (Lüscher et al., 1999). Collectively, these results strongly suggest that MCGC LTP is expressed presynaptically as a long-lasting increase in glutamate release. MC-GC LTP Requires Postsynaptic BDNF/TrkB Signaling
Author Manuscript Author Manuscript
We next sought to determine how exactly repetitive activation of MCs induces presynaptic, NMDAR-independent LTP. Group I metabotropic receptors (e.g., mGluR1/5 subtypes), which are likely activated by glutamate released during the LTP induction protocol, have been involved in some forms of presynaptic LTP (Anwyl, 2009). However, co-application of the mGluR1 antagonist LY367385 (100 μM) and the mGluR5 antagonist MPEP (4 μM) had no effect on MC-GC LTP (Figure S2A; n = 6, p > 0.3, Mann-Whitney U test), whereas in interleaved experiments, these antagonists blocked the inward current induced by the group I mGluR agonist DHPG (Figure S2B). These results indicate that MC-GC LTP does not require mGluR1/5 activation. Anatomical studies have shown a uniquely high expression of type 1 cannabinoid receptors (CB1) in the IML (Katona et al., 2006; Monory et al., 2006; Uchigashima et al., 2011), and consistent with these studies, MC-GC synapses show robust sensitivity to endocannabinoid signaling (Chiu and Castillo, 2008). Recent evidence indicates that endocannabinoids can mediate presynaptic forms of LTP whose induction requires CB1 activation both at the lateral perforant path to GC (LPP-GC) synapse in the dentate gyrus (Wang et al., 2016) and at the Schaffer collateral to CA1 pyramidal cell synapse (Gómez-Gonzalo et al., 2015). However, MC-GC LTP in the presence of the CB1 inverse agonist AM251 (4 μM) was indistinguishable from interleaved control experiments (Figure 3A; control: 182% ± 14% of baseline, n = 5, p < 0.01, paired t test; AM251: 188% ± 14% of baseline, n = 5, p < 0.01, paired t test; control versus AM251: p > 0.7, unpaired t test). As expected (Chiu and Castillo, 2008), AM251 blocked CB1-mediated depolarizationinduced suppression of excitation (DSE) at MC-GC synapses (Figure S2C; n = 4, p < 0.001, unpaired t test). These findings indicate that unlike the LTP reported at LPP-GC synapses, MC-GC LTP is a CB1-receptor-independent phenomenon.
Neuron. Author manuscript; available in PMC 2017 September 22.
Hashimotodani et al.
Page 6
Author Manuscript Author Manuscript
Another unique property of the IML is the high expression levels of brain-derived neurotrophic factor (BDNF) (Conner et al., 1997; Dieni et al., 2012; Yan et al., 1997b), a trophic factor widely involved in long-term synaptic plasticity throughout the brain (Edelmann et al., 2014; Lu et al., 2014). In the developed brain, BDNF preferentially activates TrkB receptors (Minichiello, 2009; Park and Poo, 2013), which are expressed in GCs (Donovan et al., 2008; Drake et al., 1999; Yan et al., 1997a). We therefore investigated the role of BDNF/TrkB signaling in MC-GC LTP in three ways. First, bath application of K252a (15 μM), which inhibits TrkB kinase activity, significantly reduced MC-GC LTP (Figure 3B; control: 196% ± 15% of baseline, n = 11, p < 0.001, paired t test; K252a: 116% ± 7% of baseline, n = 10, p > 0.06, paired t test; control versus K252a: p < 0.01, unpaired t test). Second, bath application of the selective TrkB receptor antagonist ANA-12 (15 μM), an agent mechanistically distinct from K252a, which prevents BDNF binding noncompetitively, abolished MC-GC LTP (Figure 3C; control: 202% ± 12% of baseline, n = 6, p < 0.001, paired t test; ANA-12: 116% ± 9% of baseline, n = 6, p > 0.1, paired t test; control versus ANA-12: p < 0.01, unpaired t test). Lastly, including the BDNF scavenger TrkB-Fc in the extracellular recording solution also significantly impaired MC-GC LTP, but not the human immunoglobulin G (IgG) that was used as control (Figure 3D; control: 203% ± 7% of baseline, n = 6, p < 0.001, paired t test; TrkB-Fc: 108% ± 21% of baseline, n = 6, p > 0.7, paired t test; control versus TrkB-Fc: p < 0.005, unpaired t test). Taken together, these three observations strongly suggest that repetitive activation of MC axons likely releases BDNF, which induces MC-GC LTP by activating TrkB receptors.
Author Manuscript
To investigate a potential role of postsynaptic TrkB receptors in MC-GC LTP, we selectively and conditionally knocked out the TrkB receptor from postsynaptic GCs by stereotaxically injecting adeno-associated virus (AAV) containing Cre recombinase (AAV5.CamKII.GFPCre) or GFP only (AAV5.CamKII.eGFP) into the dentate gyrus of adult TrkB-floxed mice. The viral vectors were strongly expressed in GCs (dorsal blade), but not in the hilus (Figure 3E), strongly suggesting that Cre-mediated recombination, and thus TrkB receptor knockout, was solely postsynaptic. We recorded from GFP+ GCs (Figure S2D) and found that MC-GC LTP was abolished in CRE-GFP+ GCs (TrkB cKO), but it was normal in GFP+ GCs (control) (Figure 3F; TrkB cKO: 90% ± 7% of baseline, n = 5, p > 0.2, paired t test; control: 158% ± 9% of baseline, n = 8, p < 0.001, paired t test; TrkB cKO versus control: p < 0.001, unpaired t test). These results indicate that postsynaptic TrkB receptors are necessary for MC-GC LTP, further supporting a role of BDNF/TrkB signaling in this form of plasticity.
Author Manuscript
To determine whether BDNF is sufficient to trigger MC-GC LTP, we bath applied recombinant human BDNF (8 nM) for 15 min. This manipulation induced long-lasting increase in MC-GC EPSC amplitude, but not MPP-GC EPSC, recorded from the same GC (Figure 4A; MC-GC: 190.9% ± 15.8% of baseline, n = 5, p < 0.01, paired t test; MPP: 101.4% ± 7.6% of baseline, n = 5, p > 0.85, paired t test; MC versus MPP: n = 5, p < 0.001, unpaired t test). Like synaptically induced LTP (Figure 2A), BDNF-induced LTP of MC-GC transmission was accompanied by a significant decrease in PPR and CV (Figure 4B), suggesting a presynaptic mechanism. Puffing BDNF (8 nM) in IML also induced longlasting potentiation of MC-GC transmission, and this potentiation was abolished in the presence of the TrkB receptor antagonist ANA-12 (15 μM) (Figure 4C; MC-GC: 149% ± 5% of baseline, n = 13, p < 0.001, paired t test; MC-GC in ANA-12: 97% ± 6% of Neuron. Author manuscript; available in PMC 2017 September 22.
Hashimotodani et al.
Page 7
Author Manuscript
baseline, n = 5, p > 0.55, paired t test). In contrast, puffing BDNF onto middle molecular layer (MML), the MPP synaptic field, had no effect on MPP-GC medial transmission (Figure 4C; MPP-GC: 100% ± 6% of baseline, n = 7, p > 0.7, paired t test). The LTP induced by BDNF puffs was also accompanied by significant decreases in PPR and CV (Figure 4D), and it was rapidly established compared with BDNF bath application (Figure 4A), presumably due to a faster effective concentration reached by the local application of BDNF. Lastly, puffing BDNF on IML increased the frequency, but not amplitude, of spontaneous EPSCs (sEPSCs) (Figures 4E and 4F) in an ANA-12-sensitive manner (Figures 4G and 4H), further supporting the idea that BDNF-induced LTP is likely due to an increase in release probability. Thus, transient activation of post-synaptic TrkB receptors by BDNF is sufficient to trigger a long-lasting increase in glutamate release at MC-GC synapses. cAMP/PKA Signaling Is Necessary and Sufficient for MC-GC LTP
Author Manuscript Author Manuscript
Presynaptically expressed forms of LTP commonly require cAMP/PKA signaling (Castillo, 2012). To test whether this metabolic cascade could also be involved in MC-GC LTP, we first examined the effect of the PKA inhibitor H89 (10 μM). We found that MC-GC LTP was abolished in the presence of this inhibitor (Figure 5A; 113% ± 12% of baseline, n = 6, p < 0.05 compared to control LTP in interleaved slices, Mann-Whitney U test). To determine the potential contribution of postsynaptic PKA activity in LTP, we included the membraneimpermeable PKA inhibitor PKI6–22 peptide (2.5 μM) in the internal recording solution. Loading GCs with PKI6–22 did not prevent MC-GC LTP (Figure 5B; 177% ± 19% of baseline, n = 5, p > 0.4 compared to control LTP in interleaved slices, Mann-Whitney U test). In contrast, intracellularly loaded PKI6–22 abolished the ability of the PKA activator Sp-cAMPS to inhibit the slow afterhyperpolarization (AHP) (IAHP) in CA1 pyramidal neurons (Pedarzani and Storm, 1993) (Figure S3A; control: 39.0% ± 5.9% of baseline, n = 5; PKI: 87.6% ± 12.6% of baseline, n = 4; control versus PKI: p < 0.05, unpaired t test). Together, these results strongly suggest that MC-GC LTP requires presynaptic, but not postsynaptic, PKA activity.
Author Manuscript
To test whether PKA activation is sufficient to trigger long-lasting potentiation of MC-GC transmission, we bath applied the adenylyl cyclase activator forskolin (FSK) for 10 min (50 μM) while monitoring MC and MPP EPSCs in the same GC. FSK elicited robust longlasting potentiation of MC-GC EPSCs (FSK-LTP) but only weak potentiation of MPP EPSCs (Figure 5C; MC: 203% ± 20% of baseline, n = 8, p < 0.005, paired t test; MPP: 136% ± 11% of baseline, n = 8, p < 0.05, paired t test). Like synaptically induced MC-GC LTP (Figure 2A), FSK-LTP of MC EPSCs was associated with significant reductions in PPR and CV (Figure 5C, right), suggesting a presynaptic expression mechanism. Consistent with this mechanism, we found that FSK had no effect on GC responses elicited by brief puffs of glutamate in the IML, a manipulation that selectively assesses postsynaptic sensitivity by shortcutting transmitter release, whereas electrically induced EPSCs in the same GC were strongly potentiated (Figure S3B). Moreover, FSK application increased the frequency, but not the amplitude, of asynchronous synaptic events recorded in the presence of strontium and elicited by stimulation in the IML (Figure S3C; n = 7, frequency: p < 0.001, Kolmogorov-Smirnov test). Collectively, these results indicate that FSK-LTP of MC-GC transmission, like synaptically induced MC-GC LTP, is likely due to an increase in
Neuron. Author manuscript; available in PMC 2017 September 22.
Hashimotodani et al.
Page 8
Author Manuscript
glutamate release probability. Lastly, we found that MC-GC LTP and FSK-LTP occlude each other: synaptically induced MC-GC LTP was significantly reduced in FSK-pretreated hippocampal slices (50 μM, 10 min) (Figure 5D; 120% ± 13% of baseline, n = 6; p < 0.01 compared to control LTP in interleaved slices, Mann-Whitney U test), and FSK application induced only a transient potentiation at MC-GC synapses expressing LTP (Figure 5E; 118% ± 14% of baseline, n = 5, p > 0.05, paired t test). The fact that MC-GC LTP and FSK-LTP occlude each other indicates that these forms of plasticity share a common mechanism, further supporting the notion that presynaptic cAMP/PKA signaling mediates MC-GC LTP.
Author Manuscript
Our results thus far indicate that both BDNF/TrkB and cAMP/PKA signaling are involved in MC-GC LTP. We therefore sought to determine potential interactions between these signaling cascades. We found that the cell-permeant PKA inhibitor myristoylated PKI14–22 peptide (1 μM) abolished BDNF-induced LTP in interleaved slices (Figure 5F; PKI14–22: 95% ± 10% of baseline, n = 4; control: 152% ± 5% of baseline, n = 6; PKI14–22 versus control: p < 0.001, unpaired t test). In contrast, FSK-LTP was indistinguishable in TrKBcKO GCs from control GFP GCs (Figure 5G; TrkB cKO: 143% ± 7% of baseline, n = 6; control: 145% ± 14% of baseline, n = 5; PKI versus control: p > 0.7, unpaired t test). These findings, together with our previous results (Figures 3, 4, 5A–5E, and S3), strongly suggest that cAMP/PKA signaling mediates MC-GC LTP downstream of BDNF/TrkB signaling. Optogenetic Activation of MCs Is Sufficient for MC-GC LTP Induction
Author Manuscript Author Manuscript
In addition to MC axons, electrical stimulation in the IML likely recruits neuromodulatory fibers (Leranth and Hajszan, 2007) and other glutamatergic axons, including ipsilateral projections from CA3 pyramidal neurons (Amaral et al., 2007). To test whether activation of MC inputs alone is sufficient for LTP induction, we employed an optogenetic approach that selectively activates MC inputs (Chancey et al., 2014; Hsu et al., 2016; Kumamoto et al., 2012). Most anatomical evidence indicates that contralateral inputs to the IML mainly, if not exclusively, arise from MCs (Amaral et al., 2007). Taking advantage of this MC commissural projection, we stereotaxically injected AAV carrying channelrhodopsin-2 with the H134R mutation AAV-hSyn-hChR2(H134R)-EYFP or AAV-hSyn-ChIEF-citrine into the ipsilateral hilus (Figure 6A), and we observed robust expression of these viral vectors in the IML of the contralateral dentate gyrus (Figures 6B–6E). We performed whole-cell voltageclamp recordings from GCs of contralateral slices and confirmed that blue light pulses delivered through the 40× immersion objective (Figure 6F) evoked tetrodotoxin (TTX)sensitive inward currents in the presence of picrotoxin (Figure 6G). These currents are optically evoked EPSCs (O-EPSCs), as indicated by their typical waveform and sensitivity to the ionotropic glutamate receptor antagonists NBQX and D-APV (Figure 6G). Like electrically evoked MC EPSCs (Figure 1), O-EPSCs were insensitive to DCG-IV (Figure 6H; 105% ± 10% of baseline, n = 5, p > 0.5, paired t test), indicating that optical stimulation spares MPP inputs. As previously reported for electrically evoked MC EPSCs (Chiu and Castillo, 2008; Monory et al., 2006), bath application of the CB1 agonist WIN 55,212-2 suppressed O-EPSCs (Figure S4A; 46.2% ± 10.3% of baseline, n = 5, p < 0.005, paired t test). In addition, as for electrically evoked MC EPSCs (Figure S2C), DSE was observed by monitoring O-EPSCs, and it was also blocked by the CB1 inverse agonist AM251 (4 μM) (Figure S4B; control: 58.1% ± 8.8% of baseline, n = 8, p < 0.01; AM251: 95.2% ± 8.2% of
Neuron. Author manuscript; available in PMC 2017 September 22.
Hashimotodani et al.
Page 9
Author Manuscript
baseline, n = 8, p > 0.05; control versus AM251: p < 0.001, unpaired t test). All these properties let us conclude that O-EPSCs recorded in GCs are indeed mediated by commissural MCs.
Author Manuscript
We next examined whether optically activated MC-GC synapses could undergo long-term plasticity. MC-GC LTP requires repetitive activation of MC axons at 30 Hz or higher (Figure 1C). However, such frequency stimulation was too fast for ChR2(H134R) to follow reliably (Figure S4C). To achieve high-fidelity optogenetic stimulation of MC axons, we employed ChIEF, a faster version of ChR2 (Lin et al., 2009). Indeed, by using ChIEF and delivering 30-Hz bursts (instead of 100 Hz), we were able to reliably activate MC axon terminals throughout the LTP induction protocol (Figure 6I, top). This induction protocol triggered robust LTP of O-EPSCs (Figure 6I, bottom; 163% ± 17% of baseline, n = 8, p < 0.01, paired t test), and like electrically induced MC-GC LTP (Figure 2A), this LTP was accompanied by a significant reduction of PPR and CV (Figure 6I, right). As for electrically activated MC axons (Figure 1F), pairing O-EPSCs with postsynaptic depolarization triggered robust NMDAR-independent MC-GC LTP, and this plasticity was also accompanied by a significant change in PPR (Figure S4D, right). Lastly, FSK application (50 μM for 10 min) induced robust LTP (Figure 6J; 219% ± 35% of baseline, n = 6, p < 0.05, Wilcoxon signed rank test) that was accompanied by a significant decrease of PPR and CV (Figure 6J, right). Altogether, our results clearly indicate that selective activation of MC axons is sufficient to induce LTP of MC-GC transmission and rule out the requirement of neuromodulatory fibers. Selective LTP in MC-GC Synapses, but Not Feed-Forward Inhibition, Increases GC Output
Author Manuscript Author Manuscript
In addition to GCs, MCs also make synapses with GABAergic interneurons that contact GCs, thereby establishing a feed-forward inhibitory circuit (Figure 1A). While the net effect of MC activity (i.e., excitation or inhibition) onto GCs remains unclear (Scharfman and Myers, 2013), MC-GC LTP could represent a dynamic way of changing the E/I balance in favor of excitation. To test this possibility, we first examined whether feed-forward inhibition could be observed by optically activating commissural MC axons (Figure 7A, left). In agreement with recent reports (Chancey et al., 2014; Hsu et al., 2016), light activation of MC axons evoked biphasic responses in the absence of picrotoxin while voltage clamping GCs at −60 mV, (Figure 7A). The outward component of this current was abolished following bath application of NBQX and D-APV, indicating its disynaptic nature (Figure 7A). In addition, the outward component was eliminated by picrotoxin (Figure 7B) or by voltage clamping GCs at −89 mV (≈ECl− in these experiments; data not shown). Thus, the most likely interpretation of these results is that optically evoked, disynaptic inhibitory postsynaptic currents (O-dIPSCs) are mediated by dentate gyrus interneurons indirectly activated by ChR2-expressing MC axons (Figure 7A, left). The scatterplot of mean amplitudes of O-EPSCs and O-dIPSCs revealed that their amplitudes are comparable under our experimental conditions (Figure 7C; O-EPSC: 44.3 ± 7.3 pA, O-dIPSC: 41.1 ± 3.9 pA). In 4 out of 28 cells, we observed a small NBQX and D-APV resistant, but picrotoxin sensitive component of O-IPSCs (Figure S5), which is presumably mediated by commissural monosynaptic inhibitory inputs from hilar interneurons (Bakst et al., 1986; Ribak et al., 1986).
Neuron. Author manuscript; available in PMC 2017 September 22.
Hashimotodani et al.
Page 10
Author Manuscript Author Manuscript
To test whether induction of MC-GC LTP could modify the E/I balance, we monitored optically induced, biphasic (excitatory→inhibitory) synaptic responses under normal conditions (i.e., in the absence of drugs) and delivered an LTP induction protocol using repetitive optical stimulation. We found that this protocol induced robust LTP of O-EPSCs, but not O-dIPSCs (Figure 7D; O-EPSC: 164% ± 22% of baseline, O-dIPSC: 89.9% ± 7.7% of baseline, n = 7; O-EPSC LTP versus O-dIPSC LTP: p < 0.01, unpaired t test). Given that the peak amplitude of O-dIPSCs was influenced by potentiated O-EPSCs, we also analyzed off-peak amplitude of O-dIPSCs (30 ms after light stimulation; Figure 7D, dotted line) but found no potentiation either (109% ± 18% of baseline, n = 7). To exclude potential contamination of commissural direct inputs from hilar interneurons (as seen in Figure S5), NBQX and D-APV were applied at the end of the experiments, and only experiments in which these antagonists abolished synaptic responses (i.e., no evidence for monosynaptic IPSCs) were included in our analysis. To directly examine whether repetitive activation of MC axons could trigger plasticity at MC-interneuron synapses, we recorded O-EPSC from contralateral dentate gyrus interneurons whose somas were in the molecular layer and at the granule cell layer/hilus border (Figure 7E). The dendritic arbor and soma location of these interneurons, a main target of the commissural projection (Hsu et al., 2016), were identified by intracellular loading of Alexa Fluor 594 (15 μM) and imaged under a two-photon microscope. These O-EPSCs were presumably monosynaptic as indicated by the short synaptic delay (3.3 ± 0.2 ms, n = 9) (Hsu et al., 2016). We found that repetitive photostimulation of MC axons (5 pulses at 30 Hz repeated 50 times at 2 Hz) failed to induce any form of long-lasting plasticity at MC-interneuron synapses (Figure 7E; 103% ± 7% of baseline, n = 9, p > 0.9, paired t test). Altogether, these results indicate that the monosynaptic excitatory MC input, but not disynaptic feed-forward inhibition, expresses LTP.
Author Manuscript Author Manuscript
Given the increase in E/I balance associated with the induction of MC-GC LTP and the close proximity of MC axon inputs to the GC soma, we predicted that selective potentiation of MC-GC synapses should increase action potential firing in GCs. To test this possibility, we monitored MC-induced burst firing in GCs before and after LTP induction. Under normal conditions (i.e., no drugs in the bath), we found that electrical, brief-burst activation of MC axons (5 pulses at 20 Hz, repeated every 20 s) failed to induce action potentials in GCs at a wide range of intensities (Figures 8A–8E). The lack of synaptically induced firing can be attributed to the strong feed-forward and tonic inhibition (Coulter and Carlson, 2007; Scharfman and Myers, 2013). In accordance with this notion, MC activation elicited action potentials in the presence of picrotoxin (Figures 8A–8C). Remarkably, after a stable 10-min baseline in which MC burst stimulation triggered no spikes, we found that the MC-GC LTP induction protocol triggered a long-lasting increase in burst firing as indicated by the appearance of one or more spikes per burst (Figures 8D–8E and 8H; spike probability; 33% ± 5%, n = 8, p < 0.05, Wilcoxon signed rank test). No significant changes in membrane input resistance and membrane potential were observed post-MC-GC LTP induction (Rinput: 105% ± 4% of baseline, n = 8, p > 0.1, paired t test; membrane potential [Vm]: baseline: 64.7 ± 0.7 mV; LTP: 64.9 ± 0.7 mV, n = 8, p > 0.5, paired t test). In addition, the long-lasting increase in burst firing was abolished when the LTP induction protocol was delivered in the presence of the PKA inhibitor H89 (10 μM) or the TrKB receptor antagonist ANA-12 (15
Neuron. Author manuscript; available in PMC 2017 September 22.
Hashimotodani et al.
Page 11
Author Manuscript
μM), two agents that also block MC-GC LTP (Figures 3C and 5A). Taken together, these results indicate that by changing the E/I balance, MC-GC LTP increases GC output.
DISCUSSION
Author Manuscript
In this study, we demonstrate that MC-GC synapses, unlike other key excitatory synapses in the dentate gyrus, undergo an NMDAR-independent and presynaptically expressed form of LTP. To our knowledge, this report is the first one to show and characterize a mechanistically unique form of long-term plasticity at identified MC-GC synapses. Unexpectedly, we found that induction of MC-GC LTP requires postsynaptic BDNF/TrkB signaling and presynaptic activation of the cAMP-PKA cascade (Figure S6). Moreover, this LTP is input specific and does not involve feed-forward inhibition mediated by MCs. As a result, selective LTP of MC-GC transmission changes the E/I balance, thereby leading to a long-lasting increase in GC output. By controlling information flow through the dentate gyrus, MC-GC LTP may contribute significantly to learning and temporal lobe epilepsy. Excitation of GCs via Monosynaptic MC-GC Synapses
Author Manuscript
The precise role of MCs in normal function and disease has remained elusive mainly because of the difficulty in selectively manipulating these neurons and monitoring their activity in behaving animals (Scharfman, 2016). Previous studies disagree on whether the net effect of MC activity on GCs is excitatory or inhibitory (Buckmaster and Schwartzkroin, 1994; Jinde et al., 2013; Ratzliff et al., 2002; Scharfman and Myers, 2013). Acute ablation of MCs decreased GC excitability in rat hippocampal slices (Ratzliff et al., 2004), whereas selective MC degeneration in mice increased dentate local field potential activity in vivo, suggesting that the net effect of MC excitation is to inhibit GC activity (Jinde et al., 2012). Our findings in acute rat and mouse hippocampal slices indicate that the net effect of MCs on GC output is dynamically regulated. Under basal conditions, feed-forward inhibition and tonic inhibition (Coulter and Carlson, 2007; Scharfman and Myers, 2013) likely impose a high threshold for GC activation, and MC inputs may not be strong enough to generate spikes in GCs. However, the selective potentiation of MC-GC transmission may provide the necessary excitatory drive to overcome inhibition.
Author Manuscript
To trigger MC-GC LTP, we used burst stimulation of MC axons with activity patterns that can occur during exploratory behaviors in rodents (Danielson et al., 2017; GoodSmith et al., 2017; Senzai and Buzsáki, 2017). An in vivo study in mouse reported that tetanization of ipsilateral PP induces LTP at the hilar associational pathway (putative GC-MC synapses), and this LTP is presumably an emergent property of the trisynaptic excitatory associative network (PP-GC-MC-GC) given that direct tetanization in IML failed to induce LTP of the associative pathway (Kleschevnikov and Routtenberg, 2003). Another in vivo study showed that induction of PP-GC LTP leads to LTP of contralateral MC-GC synapses (AlvarezSalvado et al., 2013). Using voltage imaging in hippocampal slices, it has been reported that LTP of the GC-MC-GC loop can facilitate LTP at PP-GC synapses and by this way enable information transfer from entorhinal cortex to the CA3 area (Wright and Jackson, 2014). It is therefore possible that induction of PP-GC LTP in vivo, presumably by increasing MC activity via GCs, facilitates the induction of MC-GC LTP. PP stimulation in acute slices
Neuron. Author manuscript; available in PMC 2017 September 22.
Hashimotodani et al.
Page 12
Author Manuscript
induces persistent MC activity via glutamatergic neurons in the IML known as semilunar GCs (Larimer and Strowbridge, 2010), and such activity could also contribute to the induction of MC-GC LTP in vivo.
Author Manuscript
Anatomical studies have shown that the IML of dentate gyrus receives projections from CA3 pyramidal neurons and neurons of the supramamillary nucleus of the hypothalamus (Amaral et al., 2007), and several neuromodulatory fibers terminate in the dentate gyrus as well (Leranth and Hajszan, 2007). By injecting AAV-ChR2(H134R) or -ChIEF into the hilus, we were able to successfully activate commissural MC axons with light, as recently reported by others (Chancey et al., 2014; Hsu et al., 2016; Kumamoto et al., 2012). We activated commissural MC axons in acute hippocampal slices and simultaneously monitored MCmediated monosynaptic excitation (e.g., O-EPSCs) and disynaptic inhibition (O-dIPSCs) in contralateral GCs (Figure 7). Consistent with recent observations (Chancey et al., 2014), we found comparable contribution of O-EPSCs and O-dIPSCs, supporting the notion that MCs can contribute significantly to GC excitation (Bekenstein and Lothman, 1991; Deadwyler et al., 1975; Hetherington et al., 1994; Jackson and Scharfman, 1996; Steward et al., 1977). Importantly, although electrical stimulation in the IML may recruit non-MC axons, our findings using optogenetics indicate that activation of MCs is sufficient to induce LTP. Mechanism of MC-GC LTP
Author Manuscript Author Manuscript
LTP at the hilar associational pathway to dentate gyrus inputs (e.g., GC-MC-GC) was found in an early in vivo study in rat (Hetherington et al., 1994), but not in other studies (Bekenstein and Lothman, 1991; Kleschevnikov and Routtenberg, 2003). The commissural pathway (axonal projections mainly arising from contralateral MCs) also exhibited LTP, but only when inhibition was blocked by bicuculline (Steward et al., 1990), consistent with a Hebbian form of LTP. A follow-up study demonstrated NMDAR-dependent LTP of the ipsilateral associational path (Hetherington et al., 1994). Presumably based on these early studies, it has been commonly assumed that MC-GC synapses express classical NMDARdependent, Hebbian LTP (Lisman, 1999; Wright and Jackson, 2014), although this possibility was never directly tested in vitro. Here, we report that MC-GC synapses express a presynaptic form of LTP whose induction does not require NMDAR activation. Consistent with the high expression levels of BDNF in the IML (Conner et al., 1997; Dieni et al., 2012; Yan et al., 1997b) and the presence of TrkB receptors in GCs (Donovan et al., 2008; Drake et al., 1999; Yan et al., 1997a), we found that MC-GC LTP induction requires postsynaptic BDNF-TrkB signaling as indicated by a number of complementary manipulations: molecular inactivation of endogenous BDNF, pharmacological inhibition of TrkB signaling, and TrkB receptor deletion from GCs, each impaired MC-GC LTP. Moreover, we found that BDNF is necessary and sufficient to induce plasticity. Although our results suggest that BDNF is likely secreted presynaptically from MC axons, a postsynaptic contribution from neighboring GCs or glia cannot be discarded. MC-GC LTP is likely due to a long-lasting increase in glutamate release as indicated by an associated reduction in PPR and CV and the potentiation of both AMPAR- and NMDARmediated components of synaptic transmission. Like other forms of presynaptic LTP in the brain (Castillo, 2012), we found that presynaptic activation of cAMP/PKA is necessary and
Neuron. Author manuscript; available in PMC 2017 September 22.
Hashimotodani et al.
Page 13
Author Manuscript Author Manuscript
sufficient to induce MC-GC LTP. Chemical LTP induced by forskolin was also associated with a reduction in PPR and CV and an increase in the frequency, but not amplitude, of asynchronous events, strongly suggesting a presynaptic mechanism of expression. Further, BDNF-induced LTP requires downstream PKA signaling (Figure 5F), most likely at the presynaptic terminal, but forskolin-induced LTP can occur in the absence of postsynaptic TrkB (Figure 5G). The requirement for postsynaptic TrkB signaling during induction and the presynaptic expression mechanism strongly suggests the involvement of retrograde signaling in MC-GC LTP. Although, the identity of the messenger(s) remains to be determined, it is unlikely to be released via exocytosis given that BoTx had no effect on this form of plasticity. The fact that presynaptic cAMP/PKA signaling is sufficient to trigger LTP of MCGC transmission strongly suggests that the retrograde signal targets the presynaptic cAMP/PKA cascade. Future studies will have to determine the retrograde messenger, the signaling pathway that activates presynaptic cAMP signaling and the downstream PKA targets involved in the expression of MC-GC LTP. At most excitatory synapses onto hippocampal principal cells, LTP is associative (Hebbian), NMDAR dependent, and expressed postsynaptically (Bliss et al., 2007). A good example can be found at recurrent synapses between CA3 pyramidal cells, which establish an autoassociative loop in the hippocampus (Mishra et al., 2016). In contrast, both MC-GC (this study) and GC-MC synapses express presynaptic, NMDAR-independent LTP (Lysetskiy et al., 2005). It therefore appears that activity-dependent, long-term strengthening of the intrinsic GC-MC-GC excitatory loop mainly relies on presynaptic LTP. The functional significance for this mechanistically distinct form of synaptic strengthening at the GC-MCGC loop is unclear and warrants further investigation.
Author Manuscript
Functional Relevance of MC-GC LTP Growing evidence indicates that the functional interaction between GCs and MCs is a key component of the computation performed by the dentate gyrus (Scharfman, 2016). The E/I balance shift following MC-GC LTP (Figure 7D) could potentially have a significant effect on how the dentate gyrus gates incoming information from the entorhinal cortex. Anatomically, the dentate is situated at an ideal location within the hippocampus to act as a gate, or filter, for incoming information (Hsu, 2007). This property of the dentate gyrus arises, at least in part, from the GC’s low firing rate, which creates a “sparse coding” system (Jung and McNaughton, 1993; Neunuebel and Knierim, 2012). By increasing the likelihood that a GC will fire an action potential (Figure 8), MC-GC LTP could alter the degree of sparse coding of GCs, which could in turn dynamically regulate dentate computations and memory formation processes (McNaughton and Morris, 1987).
Author Manuscript
MCs are unique among dentate gyrus neurons, because their axons extensively project along the septotemporal axis of the hippocampus (Amaral et al., 2007; Buckmaster et al., 1996), thereby connecting functionally diverse hippocampal areas (Strange et al., 2014). By increasing the connectivity between distant regions of the hippocampus, MC-GC LTP could enable or enhance computations that involve communication between septal and temporal hippocampal regions, thereby contributing to the formation of memories that require the conjunctive encoding of emotional (temporal) and spatial (septal) inputs.
Neuron. Author manuscript; available in PMC 2017 September 22.
Hashimotodani et al.
Page 14
Author Manuscript Author Manuscript
Our findings support the “granule cell association hypothesis” of MC function (Buckmaster and Schwartzkroin, 1994), in which MCs are able to excite GCs, making a positive-feedback loop (GC-MC-GC). Such a recurrent circuit may contribute to memory processing, including pattern separation (Myers and Scharfman, 2009), but may also be involved in maladaptive changes. It has been hypothesized that MCs with enhanced excitability (e.g. following a brain insult) contribute to hyperexcitability in temporal lobe epilepsy (irritable MC hypothesis) (Ratzliff et al., 2002; Santhakumar et al., 2000). By activating GCs and the associated recurrent circuits in the dentate gyrus and CA3 area, MC-GC LTP could promote seizures. Moreover, MC-GC LTP could be a mechanism by which BDNF-TrkB signaling contributes to temporal lobe epilepsy (McNamara and Scharfman, 2012). MCs could also play a role in sequence learning, where the CA3 area is an autoassociative network and the dentate gyrus is heteroassociative, involving the back projection from CA3 to dentate via MCs (Lisman, 1999; Lisman et al., 2005). In this model, the dentate and CA3 regions work together to learn and recall sequences via associative loops. The coincident timing between the associative loops and feedforward inputs from the entorhinal cortex may result in LTP induction. Given that LTP at MC-GC synapses can occur in the absence of postsynaptic activity, we predict that the coincidence detection of the heteroassociative loop mainly relies on the CA3-MC connection. Future studies will have to investigate how exactly LTP at MCGC synapses contribute to dentate-gyrus-dependent learning, including pattern separation, sequence learning, and epilepsy.
STAR★METHODS Detailed methods are provided in the online version of this paper and include the following:
Author Manuscript
KEY RESOURCES TABLE
REAGENT or RESOURCE
Author Manuscript
SOURCE
IDENTIFIER
NBQX
Abcam
Cat#ab120046
NBQX
Chemical Synthesis and Drug Supply Program of the National Institute of Mental Health
N/A
D-APV
Abcam
Cat#ab120003
D-APV
Chemical Synthesis and Drug Supply Program of the National Institute of Mental Health
N/A
Forskolin
Abcam
Cat#ab120058
AM251
Abcam
Cat#ab120088
Picrotoxin
Sigma-Aldrich
Cat#P1675
H89
Sigma-Aldrich
Cat#B1427
K252a
Sigma-Aldrich
Cat#82497
PKI6–22
Enzo Life Science
Cat# BML-P204-0001
Sp-cAMPs
Enzo Life Science
Cat# BML-CN136-0001
TrkB-Fc
R&D systems
Cat#688-TK
Chemicals, Peptides, and Recombinant Proteins
Neuron. Author manuscript; available in PMC 2017 September 22.
Hashimotodani et al.
Page 15
Author Manuscript Author Manuscript
REAGENT or RESOURCE
SOURCE
IDENTIFIER
Human IgG
R&D systems
Cat#1-001-A
Botulinum toxin-B
List biological
Cat# 620A
BDNF (human)
Tocris Bioscience
Cat#2837
DCG-IV
Tocris Bioscience
Cat#0975
ANA-12
Tocris Bioscience
Cat#4781
Glutamate (L-glutamic acid)
Tocris Bioscience
Cat#2018
PKI14–22
Tocris Bioscience
Cat#2546
Tetrodotoxin
Tocris Bioscience
Cat#1078
CGP55845
Tocris Bioscience
Cat#1248
MPEP
Tocris Bioscience
Cat#1212
LY367385
Tocris Bioscience
Cat#1237
(S)-3,5-DHPG
Tocris Bioscience
Cat#0805
WIN55,212-2
Tocris Bioscience
Cat#1038
Alexa Fluor 594 Hydrazide
Invitrogen Molecular Probes
Cat#A10438
Rat: Sprague-Dawley
Charles River
Cat# 400
Mouse: C57BL/6NCrl
Charles River
Cat# 027
Mouse: TrkB floxed
Dr. Lisa Monteggia
N/A
AAV1-hSyn-ChR(H134R)-eYFP
Penn Vector Core
AV-1-26973P
AAV2/1-hSyn-ChIEF-Citrine
Dr. Yoav Ben-Simon
N/A
AAV5-CamKII-GFP-Cre
Penn Vector Core
AV-5-PV2521
AAV5-CamKII-eGFPd
Penn Vector Core
AV-5-PV1917
IgorPro
Wavemetrics
https://www.wavemetrics.com/
OriginPro 9
OriginLab
http://www.originlab.com/
ImageJ
ImageJ
http://imagej.net/Welcome
Experimental Models: Organisms/Strains
Recombinant DNA
Author Manuscript
Software and Algorithms
CONTACT FOR REAGENT AND RESOURCE SHARING Further information and requests for resources and reagents should be directed to and will be fulfilled by the Lead Contact, Pablo E. Castillo (
[email protected]). EXPERIMENTAL MODEL AND SUBJECT DETAILS
Author Manuscript
Postnatal day 19 (P19) to P29 Sprague-Dawley rats and C57BL/6 mice of either sex, or P56–P79 TrkB floxed (Trkbflox/flox) mice of either sex were used for electrophysiological experiments. All animals were group housed in a standard 12 hr light/12 hr dark cycle. Animal handling and use followed a protocol approved by the Animal Care and Use Committee of Albert Einstein College of Medicine, in accordance with the National Institutes of Health guidelines. TrkB floxed mice, generated by Dr. Luis Parada, were kindly donated by Dr. Lisa Monteggia (University of Texas, Southwestern Medical Center).
Neuron. Author manuscript; available in PMC 2017 September 22.
Hashimotodani et al.
Page 16
METHOD DETAILS
Author Manuscript Author Manuscript
Hippocampal slice preparation—Acute transverse hippocampal slices (300–400 μm thick) were prepared from Sprague-Dawley rats, C57BL/6 mice, and TrkB floxed mice (300 μm thick). Briefly, the hippocampi were isolated and cut using a VT1200s microsclicer (Leica Microsystems Co.) or a DTK-2000 vibrating microslicer (Dosaka EM) in a solution containing (in mM): 215 sucrose, 2.5 KCl, 26 NaHCO3, 1.6 NaH2PO4, 1 CaCl2, 4 MgCl2, 4 MgSO4 and 20 D-glucose. At 30 min post sectioning, the cutting medium was gradually switched to extracellular artificial cerebrospinal (ACSF) recording solution containing (in mM): 124 NaCl, 2.5 KCl, 26 NaHCO3, 1 NaH2PO4, 2.5 CaCl2, 1.3 MgSO4 and 10 Dglucose. Slices were incubated for at least 30 min in the ACSF solution before recording. Hippocampal slices from TrkB floxed mice were prepared using a NMDG-based cutting solution containing (in mM): 93 N-Methyl-d-glucamin, 2.5 KCl, 1.25 NaH2PO4, 30 NaHCO3, 20 HEPES, 25 D-glucose, 2 Thiourea, 5 Na-Ascorbate, 3 Na-Pyruvate, 0.5 CaCl2, 10 MgCl2. These slices were then transferred to 32°C ACSF for 30 min and then kept at room temperature for at least 1h before recording. All solutions were equilibrated with 95% O2 and 5% CO2 (pH 7.4).
Author Manuscript
Electrophysiology—All experiments, unless otherwise stated, were performed at 28 ± 1°C in a submersion-type recording chamber perfused at ~2 mL min−1 with ACSF supplemented with the GABAA receptor antagonist picrotoxin (100 μM). Whole-cell patchclamp recordings using a Multiclamp 700A amplifier (Molecular Devices) were made from GCs voltage clamped at −60 mV (unless otherwise stated) using patch-type pipette electrodes (~3–4 MΩ) containing (in mM): 131 cesium gluconate, 8 NaCl, 1 CaCl2, 10 EGTA, 10 D-glucose and 10 HEPES, pH 7.2 (285–290 mOsm). For experiments using a K+based internal solution, the following intracellular solution was used (in mM): 135 KMeSO4, 5 KCl, 1 CaCl2, 5 NaOH, 10 HEPES, 5 MgATP, 0.4 Na3GTP, 5 EGTA and 10 D-glucose, pH 7.2 (280–290 mOsm). Series resistance (~7–22 MΩ) was monitored throughout all experiments with a −5 mV, 80 ms voltage step, and cells that exhibited a significant change in series resistance (> 20%) were excluded from analysis.
Author Manuscript
Stimulating patch-type pipettes were filled with ACSF and placed in the IML (< 40 μm from the border of the GC body layer) to activate MC axons, and in the middle third of the molecular layer to activate MPP inputs. To elicit synaptic responses, paired, monopolar square-wave voltage or current pulses (100–200 μs pulse width, 4–25 V) were delivered through a stimulus isolator (Isoflex, AMPI, or Digitimer DS2A-MKII) connected to a broken tip (~10–20 μm) stimulating patch-type micropipette filled with ACSF. Typically, stimulation (both electrical and optical) was adjusted to obtain comparable magnitude synaptic responses across experiments; e.g., 50–100 pA EPSCs (Vh −60 mV), and 5 mV EPSPs (Figure 8). Stimulation intensity in voltage-clamp and current-clam experiments were comparable. For minimal stimulation (Figures 2C and 2D), stimulating pipettes were made from theta glass capillaries. To confirm the specificity of stimulations, at the end of experiments we routinely applied the mGluR2/3 agonist DCG-IV (1 μM), which selectively reduces MPP-GC, but not MC-GC synaptic transmission (Chiu and Castillo, 2008). MC-GC LTP was typically induced with brief bursts (5 pulses, 100 Hz) repeated 50 times at 2 Hz, while postsynaptic neurons were continuously voltage-clamped at −60 mV. When a K+-
Neuron. Author manuscript; available in PMC 2017 September 22.
Hashimotodani et al.
Page 17
Author Manuscript Author Manuscript
based internal solution was used, the recording configuration was switched from voltageclamp to current-clamp mode during the LTP induction protocol. NMDAR-dependent (Hebbian) LTP (Figures 1F, 1G, and S4D) was induced by a pairing protocol (presynaptic stimulation + postsynaptic depolarization) consisting of 200 pulses (extracellular stimulation) at 2 Hz, paired with 100 s depolarization from −60 to 0 mV. Strontium-induced asynchronous synaptic responses were evoked every 10 s in ACSF supplemented with picrotoxin (100 μM) and extracellular CaCl2 was replaced with 8 mM SrCl2. The amplitude and frequency of the asynchronous events were measured during a 700 ms period beginning 30 ms after stimulus in order to exclude the initial synchronous synaptic responses. IAHP was elicited by a 60–70 mV, 100 ms depolarizing step in CA1 pyramidal cells voltage clamped between −50 to −55 mV with an intracellular solution containing (in mM): 135 Kgluconate, 5 KCl, 1 CaCl2, 10 HEPES, 10 D-glucose, 5 Mg-ATP, 0.4 Na3GTP, and 5 EGTA. Extracellular field recordings (fEPSPs) were performed using patch-type pipettes filled with 1M NaCl and placed in the IML. BoTx experiments were performed at 32°C. BoTx was supplemented with 5 mM dithiothreitol (DTT) in the intracellular solution, and interleaved control experiments included DTT only.
Author Manuscript
GC firing (Figure 8) was assessed using burst stimulation in the IML (5 pulses at 20 Hz every 20 s), while the membrane potential was held at −65 mV. For these experiments, the intracellular solution contained (in mM): 140 K-gluconate, 4 NaCl, 15 HEPES, 4 Mg-ATP, 0.3 Na3GTP and 0.2 EGTA, pH 7.2 (285 mOsm). Exogenous glutamate-evoked responses (Figure S3B) were evoked by directly puffing glutamate (1 mM, 20 ms, 2 PSI) using a Picospritzer III (Parker) connected to a patch pipette (resistance, 3–4 MΩ). The tip of the puffer pipette was positioned in the IML < 30 μm deep from the surface of the slice and as close as possible to the recorded cell. For BDNF puffs (8 nM, 2.5–3 PSI, 3 s puffs repeated twice, 5 s interval), the tip of the puffer pipette was positioned above the IML or MML while monitoring MC-GC or MPP-GC transmission, respectively. For interneuron whole cell recordings (Figure 7), 15 μM Alexa Fluor 594 Hydrazide was added to the internal solution and the dendritic tree was imaged using a two-photon laser scanning fluorescence microscope (60× fluor objective, NA = 1.00, Nikon) equipped with an Insight Ti:Sapphire laser (Spectra Physics) tuned to 820 nm and a galvanometer-based scanhead. Laser power at the back aperture was between 5.6 and 31.7 mW. Images were created on the NI 6110 board acquired with Prairie View 5.3U2 Beta software. ImageJ was used for analysis.
Author Manuscript
Optogenetics—Rats (P9–P11) were placed in a stereotaxic frame, and anesthetized with isoflurane (up to 5% for induction and 1%–3% for maintenance). AAV-hSyn-ChR2(H134R)EYFP (University of Pennsylvania Vector Core) or AAV-hSyn-ChIEF-Citrine (kindly provided from Dr. Y. Ben-Simon, Tel-Aviv University, Israel) was injected into the ipsilateral dentate gyrus (3.4 mm posterior to bregma, 3.2 mm lateral to bregma, 3.2 mm ventral from dura) in a total volume of 2.0 μL at a flow rate of 0.2 μl/min. In vitro electrophysiological recordings were performed 10–16 days after injection from transverse hippocampal slices obtained from contralateral hippocampus. Pulses of blue light (0.1–5.0 ms) were provided with a 473 nm laser (OEM laser systems, Inc) attached with an optic fiber (200 μm diameter), collimated and delivered through the microscope objective (40×, 0.8 NA). A light pulse was centered in the IML (for Figure 7A).
Neuron. Author manuscript; available in PMC 2017 September 22.
Hashimotodani et al.
Page 18
Author Manuscript
TrkB conditional postsynaptic KO—TrkB floxed (Trkbflox/flox) mice (4-week old) were placed in a stereotaxic frame, and anesthetized with isoflurane (up to 5% for induction and 1%–3% for maintenance) and either AAV5.CamKII.eGFP or AAV5.CamKII.GFP-CRE (University of Pennsylvania Vector Core) was injected (1.5 μL at a flow rate of 0.1 μl/min) unilaterally into the ipsilateral dorsal blade of the dentate gyrus (2.06 mm posterior to bregma, 1.5 mm lateral to bregma, 1.8 mm ventral from dura), using a beveled needle. Both male and female mice were used with a similar ratio for the two types of viruses. Slices for electrophysiology were prepared from injected animals 4–7 weeks after injection. For each animal, we verified the absence of GFP-expressing cells in the hilus of the whole ipsilateral hippocampus. Some slices were fixed in 4% PFA, stained with DAPI and single plane confocal images were collected on a Zeiss LSM 510 Meta Duo V2 confocal microscope (63× plan apochromat objective, NA = 1.4).
Author Manuscript Author Manuscript
Data analysis—Electrophysiological data were acquired at 5 kHz, filtered at 2.4 kHz, and analyzed using custom-made software for IgorPro (Wavemetrics Inc.). The magnitude of LTP was determined by comparing 10 min baseline responses with responses 20–30 min (or 30–40 min for Figures 1B, 5B, 5D, S1A, and S1B) after LTP induction. To avoid wash-out in the pairing protocol experiments, a baseline of only a 4–5 min baseline was obtained. PPR was defined as the ratio of the amplitude of the second EPSC (baseline taken 1–2 ms before the stimulus artifact) to the amplitude of the first EPSC. CV was calculated as the standard deviation of EPSC amplitude divided by mean EPSC amplitude. Both PPR and CV were measured 10 min before and 30–40 min (or 20–30 min for Figure 6) after LTP induction protocol or forskolin-induced potentiation. Synaptic delay was calculated as the time elapsed from the onset of photostimulation to the onset of the O-EPSC. Spike probability was calculated by the number of spikes normalized to the total number of spikes per burst at 25– 30 min after LTP induction. Averaged traces include 15–30 consecutive individual responses. Pharmacology—Reagents were bath applied following dilution into ACSF from stock solutions stored at −20°C prepared in water or DMSO, depending on the manufacturer’s recommendation. The final DMSO concentration was < 0.01% total volume. For experiments requiring postsynaptic loading PKI6–22, LTP was induced at least 20 min after establishing whole-cell configuration. QUANTIFICATION AND STATISTICAL ANALYSIS
Author Manuscript
Statistical analysis was performed using OriginPro software (OriginLab). The normality of distributions was assessed using the Shapiro-Wilk test. In normal distributions, Student’s unpaired and paired two-tailed t tests were used to assess between-group and within-group differences, respectively. The non-parametric paired sample Wilcoxon signed rank test and Mann-Whitney’s U test were used in non-normal distributions. Statistical significance was set to p < 0.05 (*** indicates p < 0.001, ** indicates p < 0.01, and * indicates p < 0.05). Cumulative probability plots of asynchronous MC EPSCs were compared using the Kolmogorov-Smirnov test. All values are reported as the mean ± SEM. Statistical results are summarized in Table S1.
Neuron. Author manuscript; available in PMC 2017 September 22.
Hashimotodani et al.
Page 19
Author Manuscript
Supplementary Material Refer to Web version on PubMed Central for supplementary material.
Acknowledgments
Author Manuscript
We thank Dr. Yoav Ben-Simon (Tel Aviv University) for providing AAV-hSyn-ChIEF-citrine, Dr. Lisa Monteggia (University of Texas, Southwestern Medical Center) for providing TrkB-floxed mice, and Hannah Monday for critical review of the manuscript and assistance with confocal imaging. We also thank Stefano Lutzu and Pablo Lituma for their help with the two-photon laser microscope. This work was supported by NIH R01 grants DA017392 and MH081935 to P.E.C. Y.H. was supported by the Japan Society for the Promotion of Science postdoctoral fellowships for research abroad (H23.829). K.N. was supported by the Fondation pour la Recherche Médicale (postdoctoral fellowship for research abroad) and the Fondation Bettencourt Schueller (Prix pour les jeunes chercheurs 2016). A.E.C. was partially supported by a Ruth L. Kirschstein Award (F32 NS071821), a NARSAD Young Investigator Grant from the Brain & Behavior Research Foundation, and by the Millennium Nucleus Nu-MIND (NC 130011). D.C. was supported by the Maximizing Access to Research Careers Program (T34 GM008718) and the Albert Einstein College of Medicine Summer Undergraduate Research Program (R25 GM104547).
References
Author Manuscript Author Manuscript
Alvarez-Salvado E, Pallarés V, Moreno A, Canals S. Functional MRI of long-term potentiation: imaging network plasticity. Philos Trans R Soc Lond B Biol Sci. 2013; 369:20130152. [PubMed: 24298154] Amaral DG, Witter MP. The three-dimensional organization of the hippocampal formation: a review of anatomical data. Neuroscience. 1989; 31:571–591. [PubMed: 2687721] Amaral DG, Scharfman HE, Lavenex P. The dentate gyrus: fundamental neuroanatomical organization (dentate gyrus for dummies). Prog Brain Res. 2007; 163:3–22. [PubMed: 17765709] Anwyl R. Metabotropic glutamate receptor-dependent long-term potentiation. Neuropharmacology. 2009; 56:735–740. [PubMed: 19705571] Bakst I, Avendano C, Morrison JH, Amaral DG. An experimental analysis of the origins of somatostatin-like immunoreactivity in the dentate gyrus of the rat. J Neurosci. 1986; 6:1452–1462. [PubMed: 2872280] Bekenstein JW, Lothman EW. Electrophysiological characterization of associational pathway terminating on dentate gyrus granule cells in the rat. Hippocampus. 1991; 1:399–404. [PubMed: 1669318] Bliss, TV., Collingridge, GL., Morris, RG. Synaptic plasticity in the hippocampus. In: Andersen, PA., editor. The Hippocampus Book. Oxford University Press; 2007. p. 343-474. Buckmaster PS, Schwartzkroin PA. Hippocampal mossy cell function: a speculative view. Hippocampus. 1994; 4:393–402. [PubMed: 7874231] Buckmaster PS, Wenzel HJ, Kunkel DD, Schwartzkroin PA. Axon arbors and synaptic connections of hippocampal mossy cells in the rat in vivo. J Comp Neurol. 1996; 366:271–292. [PubMed: 8698887] Castillo PE. Presynaptic LTP and LTD of excitatory and inhibitory synapses. Cold Spring Harb Perspect Biol. 2012; 4:a005728. [PubMed: 22147943] Chancey JH, Poulsen DJ, Wadiche JI, Overstreet-Wadiche L. Hilar mossy cells provide the first glutamatergic synapses to adult-born dentate granule cells. J Neurosci. 2014; 34:2349–2354. [PubMed: 24501373] Chiu CQ, Castillo PE. Input-specific plasticity at excitatory synapses mediated by endocannabinoids in the dentate gyrus. Neuropharmacology. 2008; 54:68–78. [PubMed: 17706254] Conner JM, Lauterborn JC, Yan Q, Gall CM, Varon S. Distribution of brain-derived neurotrophic factor (BDNF) protein and mRNA in the normal adult rat CNS: evidence for anterograde axonal transport. J Neurosci. 1997; 17:2295–2313. [PubMed: 9065491] Coulter DA, Carlson GC. Functional regulation of the dentate gyrus by GABA-mediated inhibition. Prog Brain Res. 2007; 163:235–243. [PubMed: 17765722]
Neuron. Author manuscript; available in PMC 2017 September 22.
Hashimotodani et al.
Page 20
Author Manuscript Author Manuscript Author Manuscript Author Manuscript
Danielson NB, Turi GF, Ladow M, Chavlis S, Petrantonakis PC, Poirazi P, Losonczy A. In vivo imaging of dentate gyrus mossy cells in behaving mice. Neuron. 2017; 93:552–559.e4. [PubMed: 28132825] Deadwyler SA, West JR, Cotman CW, Lynch GS. A neurophysiological analysis of commissural projections to dentate gyrus of the rat. J Neurophysiol. 1975; 38:167–184. [PubMed: 162942] Dieni S, Matsumoto T, Dekkers M, Rauskolb S, Ionescu MS, Deogracias R, Gundelfinger ED, Kojima M, Nestel S, Frotscher M, Barde YA. BDNF and its pro-peptide are stored in presynaptic dense core vesicles in brain neurons. J Cell Biol. 2012; 196:775–788. [PubMed: 22412021] Donovan MH, Yamaguchi M, Eisch AJ. Dynamic expression of TrkB receptor protein on proliferating and maturing cells in the adult mouse dentate gyrus. Hippocampus. 2008; 18:435–439. [PubMed: 18240316] Drake CT, Milner TA, Patterson SL. Ultrastructural localization of full-length trkB immunoreactivity in rat hippocampus suggests multiple roles in modulating activity-dependent synaptic plasticity. J Neurosci. 1999; 19:8009–8026. [PubMed: 10479701] Edelmann E, Lessmann V, Brigadski T. Pre- and postsynaptic twists in BDNF secretion and action in synaptic plasticity. Neuropharmacology. 2014; 76:610–627. [PubMed: 23791959] Fanselow MS, Dong HW. Are the dorsal and ventral hippocampus functionally distinct structures? Neuron. 2010; 65:7–19. [PubMed: 20152109] Gómez-Gonzalo M, Navarrete M, Perea G, Covelo A, Martín-Fernández M, Shigemoto R, Luján R, Araque A. Endocannabinoids induce lateral long-term potentiation of transmitter release by stimulation of gliotransmission. Cereb Cortex. 2015; 25:3699–3712. [PubMed: 25260706] GoodSmith D, Chen X, Wang C, Kim SH, Song H, Burgalossi A, Christian KM, Knierim JJ. Spatial representations of granule cells and mossy cells of the dentate gyrus. Neuron. 2017; 93:677–690. [PubMed: 28132828] Hetherington PA, Austin KB, Shapiro ML. Ipsilateral associational pathway in the dentate gyrus: an excitatory feedback system that supports N-methyl-D-aspartate-dependent long-term potentiation. Hippocampus. 1994; 4:422–438. [PubMed: 7874234] Hsu D. The dentate gyrus as a filter or gate: a look back and a look ahead. Prog Brain Res. 2007; 163:601–613. [PubMed: 17765740] Hsu TT, Lee CT, Tai MH, Lien CC. Differential recruitment of dentate gyrus interneuron types by commissural versus perforant pathways. Cereb Cortex. 2016; 26:2715–2727. [PubMed: 26045570] Jackson MB, Scharfman HE. Positive feedback from hilar mossy cells to granule cells in the dentate gyrus revealed by voltage-sensitive dye and microelectrode recording. J Neurophysiol. 1996; 76:601–616. [PubMed: 8836247] Jinde S, Zsiros V, Jiang Z, Nakao K, Pickel J, Kohno K, Belforte JE, Nakazawa K. Hilar mossy cell degeneration causes transient dentate granule cell hyperexcitability and impaired pattern separation. Neuron. 2012; 76:1189–1200. [PubMed: 23259953] Jinde S, Zsiros V, Nakazawa K. Hilar mossy cell circuitry controlling dentate granule cell excitability. Front Neural Circuits. 2013; 7:14. [PubMed: 23407806] Jung MW, McNaughton BL. Spatial selectivity of unit activity in the hippocampal granular layer. Hippocampus. 1993; 3:165–182. Katona I, Urbán GM, Wallace M, Ledent C, Jung KM, Piomelli D, Mackie K, Freund TF. Molecular composition of the endocannabinoid system at glutamatergic synapses. J Neurosci. 2006; 26:5628–5637. [PubMed: 16723519] Kesner RP, Rolls ET. A computational theory of hippocampal function, and tests of the theory: new developments. Neurosci Biobehav Rev. 2015; 48:92–147. [PubMed: 25446947] Kleschevnikov AM, Routtenberg A. Long-term potentiation recruits a trisynaptic excitatory associative network within the mouse dentate gyrus. Eur J Neurosci. 2003; 17:2690–2702. [PubMed: 12823476] Knierim JJ, Neunuebel JP. Tracking the flow of hippocampal computation: Pattern separation, pattern completion, and attractor dynamics. Neurobiol Learn Mem. 2016; 129:38–49. [PubMed: 26514299]
Neuron. Author manuscript; available in PMC 2017 September 22.
Hashimotodani et al.
Page 21
Author Manuscript Author Manuscript Author Manuscript Author Manuscript
Kumamoto N, Gu Y, Wang J, Janoschka S, Takemaru K, Levine J, Ge S. A role for primary cilia in glutamatergic synaptic integration of adult-born neurons. Nat Neurosci. 2012; 15:399–405. [PubMed: 22306608] Larimer P, Strowbridge BW. Nonrandom local circuits in the dentate gyrus. J Neurosci. 2008; 28:12212–12223. [PubMed: 19020015] Larimer P, Strowbridge BW. Representing information in cell assemblies: persistent activity mediated by semilunar granule cells. Nat Neurosci. 2010; 13:213–222. [PubMed: 20037579] Leranth C, Hajszan T. Extrinsic afferent systems to the dentate gyrus. Prog Brain Res. 2007; 163:63– 84. [PubMed: 17765712] Lin JY, Lin MZ, Steinbach P, Tsien RY. Characterization of engineered channelrhodopsin variants with improved properties and kinetics. Biophys J. 2009; 96:1803–1814. [PubMed: 19254539] Lisman JE. Relating hippocampal circuitry to function: recall of memory sequences by reciprocal dentate-CA3 interactions. Neuron. 1999; 22:233–242. [PubMed: 10069330] Lisman JE, Talamini LM, Raffone A. Recall of memory sequences by interaction of the dentate and CA3: a revised model of the phase precession. Neural Netw. 2005; 18:1191–1201. [PubMed: 16233972] Lledo PM, Zhang X, Südhof TC, Malenka RC, Nicoll RA. Postsynaptic membrane fusion and longterm potentiation. Science. 1998; 279:399–403. [PubMed: 9430593] Lorente De Nó R. Studies on the structure of the cerebral cortex. Continuation of the study of the ammonic system. J Psychol Neurol. 1934; 46:113–177. Lu B, Nagappan G, Lu Y. BDNF and synaptic plasticity, cognitive function, and dysfunction. Handb Exp Pharmacol. 2014; 220:223–250. [PubMed: 24668475] Lüscher C, Xia H, Beattie EC, Carroll RC, von Zastrow M, Malenka RC, Nicoll RA. Role of AMPA receptor cycling in synaptic transmission and plasticity. Neuron. 1999; 24:649–658. [PubMed: 10595516] Lysetskiy M, Földy C, Soltesz I. Long- and short-term plasticity at mossy fiber synapses on mossy cells in the rat dentate gyrus. Hippocampus. 2005; 15:691–696. [PubMed: 15986406] Macek TA, Winder DG, Gereau RW 4th, Ladd CO, Conn PJ. Differential involvement of group II and group III mGluRs as autoreceptors at lateral and medial perforant path synapses. J Neurophysiol. 1996; 76:3798–3806. [PubMed: 8985877] McNamara, JO., Scharfman, HE. Temporal lobe epilepsy and the BDNF receptor, TrkB In Jasper’s Basic Mechanisms of the Epilepsies. Noebels, JL.Avoli, M.Rogawski, MA.Olsen, RW., DelgadoEscueta, AV., editors. Oxford University Press; 2012. p. 514-531. McNaughton BL, Morris RGM. Hippocampal synaptic enhancement and information storage within a distributed memory system. Trends Neurosci. 1987; 10:408–415. Minichiello L. TrkB signalling pathways in LTP and learning. Nat Rev Neurosci. 2009; 10:850–860. [PubMed: 19927149] Mishra RK, Kim S, Guzman SJ, Jonas P. Symmetric spike timing-dependent plasticity at CA3-CA3 synapses optimizes storage and recall in autoassociative networks. Nat Commun. 2016; 7:11552. [PubMed: 27174042] Monory K, Massa F, Egertová M, Eder M, Blaudzun H, Westenbroek R, Kelsch W, Jacob W, Marsch R, Ekker M, et al. The endocannabinoid system controls key epileptogenic circuits in the hippocampus. Neuron. 2006; 51:455–466. [PubMed: 16908411] Myers CE, Scharfman HE. A role for hilar cells in pattern separation in the dentate gyrus: a computational approach. Hippocampus. 2009; 19:321–337. [PubMed: 18958849] Neunuebel JP, Knierim JJ. Spatial firing correlates of physiologically distinct cell types of the rat dentate gyrus. J Neurosci. 2012; 32:3848–3858. [PubMed: 22423105] Park H, Poo MM. Neurotrophin regulation of neural circuit development and function. Nat Rev Neurosci. 2013; 14:7–23. [PubMed: 23254191] Pedarzani P, Storm JF. PKA mediates the effects of monoamine transmitters on the K+ current underlying the slow spike frequency adaptation in hippocampal neurons. Neuron. 1993; 11:1023– 1035. [PubMed: 8274274] Ramón y Cajal S. Histologie du Système Nerveux de l’Homme et des Vertébrés (Malione). 1911
Neuron. Author manuscript; available in PMC 2017 September 22.
Hashimotodani et al.
Page 22
Author Manuscript Author Manuscript Author Manuscript Author Manuscript
Ratzliff, Ad, Santhakumar, V., Howard, A., Soltesz, I. Mossy cells in epilepsy: rigor mortis or vigor mortis? Trends Neurosci. 2002; 25:140–144. [PubMed: 11852145] Ratzliff, Ad, Howard, AL., Santhakumar, V., Osapay, I., Soltesz, I. Rapid deletion of mossy cells does not result in a hyperexcitable dentate gyrus: implications for epileptogenesis. J Neurosci. 2004; 24:2259–2269. [PubMed: 14999076] Ribak CE, Seress L, Peterson GM, Seroogy KB, Fallon JH, Schmued LC. A GABAergic inhibitory component within the hippocampal commissural pathway. J Neurosci. 1986; 6:3492–3498. [PubMed: 2432200] Santhakumar V, Bender R, Frotscher M, Ross ST, Hollrigel GS, Toth Z, Soltesz I. Granule cell hyperexcitability in the early posttraumatic rat dentate gyrus: the ‘irritable mossy cell’ hypothesis. J Physiol. 2000; 524:117–134. [PubMed: 10747187] Scharfman HE. Electrophysiological evidence that dentate hilar mossy cells are excitatory and innervate both granule cells and interneurons. J Neurophysiol. 1995; 74:179–194. [PubMed: 7472322] Scharfman HE. The enigmatic mossy cell of the dentate gyrus. Nat Rev Neurosci. 2016; 17:562–575. [PubMed: 27466143] Scharfman HE, Myers CE. Hilar mossy cells of the dentate gyrus: a historical perspective. Front Neural Circuits. 2013; 6:106. [PubMed: 23420672] Senzai Y, Buzsáki G. Physiological properties and behavioral correlates of hippocampal granule cells and mossy cells. Neuron. 2017; 93:691–704. [PubMed: 28132824] Sloviter RS. Permanently altered hippocampal structure, excitability, and inhibition after experimental status epilepticus in the rat: the “dormant basket cell” hypothesis and its possible relevance to temporal lobe epilepsy. Hippocampus. 1991; 1:41–66. [PubMed: 1688284] Steward O, White WF, Cotman CW. Potentiation of the excitatory synaptic action of commissural, associational and entorhinal afferents to dentate granule cells. Brain Res. 1977; 134:551–560. [PubMed: 198066] Steward O, Tomasulo R, Levy WB. Blockade of inhibition in a pathway with dual excitatory and inhibitory action unmasks a capability for LTP that is otherwise not expressed. Brain Res. 1990; 516:292–300. [PubMed: 2364294] Strange BA, Witter MP, Lein ES, Moser EI. Functional organization of the hippocampal longitudinal axis. Nat Rev Neurosci. 2014; 15:655–669. [PubMed: 25234264] Uchigashima M, Yamazaki M, Yamasaki M, Tanimura A, Sakimura K, Kano M, Watanabe M. Molecular and morphological configuration for 2-arachidonoylglycerol-mediated retrograde signaling at mossy cell-granule cell synapses in the dentate gyrus. J Neurosci. 2011; 31:7700– 7714. [PubMed: 21613483] Wang W, Trieu BH, Palmer LC, Jia Y, Pham DT, Jung KM, Karsten CA, Merrill CB, Mackie K, Gall CM, et al. A primary cortical input to hippocampus expresses a pathway-specific and endocannabinoid-dependent form of long-term potentiation. eNeuro. 2016; 3 Wright BJ, Jackson MB. Long-term potentiation in hilar circuitry modulates gating by the dentate gyrus. J Neurosci. 2014; 34:9743–9753. [PubMed: 25031412] Yan Q, Radeke MJ, Matheson CR, Talvenheimo J, Welcher AA, Feinstein SC. Immunocytochemical localization of TrkB in the central nervous system of the adult rat. J Comp Neurol. 1997a; 378:135–157. [PubMed: 9120052] Yan Q, Rosenfeld RD, Matheson CR, Hawkins N, Lopez OT, Bennett L, Welcher AA. Expression of brain-derived neurotrophic factor protein in the adult rat central nervous system. Neuroscience. 1997b; 78:431–448. [PubMed: 9145800]
Neuron. Author manuscript; available in PMC 2017 September 22.
Hashimotodani et al.
Page 23
Author Manuscript
Highlights •
Mossy cell (MC) to dentate granule cell (GC) synapses undergo robust LTP in rodents
•
MC-GC LTP is NMDA receptor independent and expressed presynaptically
•
MC-GC LTP involves postsynaptic BDNF/TrkB and presynaptic PKA signaling
•
By increasing E/I balance, MC-GC LTP enhances dentate gyrus output
Author Manuscript Author Manuscript Author Manuscript Neuron. Author manuscript; available in PMC 2017 September 22.
Hashimotodani et al.
Page 24
Author Manuscript Author Manuscript Author Manuscript
Figure 1. Repetitive Activation of MC Axons Induces NMDAR-Independent LTP at MC-GC Synapses
Author Manuscript
(A) Top: schematic diagram illustrating the neural circuit in dentate gyrus and the recording configuration. Whole-cell patch-clamp recordings were performed from GCs, and the stimulation electrode was placed in the IML to stimulate MC axons. The LTP induction protocol is depicted on the bottom. OML, outer molecular layer; MML, middle molecular layer; IML, inner molecular layer; GCL, granule cell layer; LPP, lateral perforant path; MPP, medial perforant path; GC, granule cell; IN, interneuron; MC, mossy cell. (B) MC-GC LTP representative experiment (top) and summary plot (bottom). MC EPSCs were recorded from GCs in whole-cell voltage-clamp mode. After a stable (~10-min) baseline, the induction protocol was delivered at the time indicated by the vertical arrow. At the end of the experiments, DCG-IV (1 μM) was bath applied in order to verify the identity of the evoked EPSCs. Representative traces, which correspond to the numbers in the timecourse plot below (for this and all subsequent figures), are shown on the top. (C and D) The magnitude of MC-GC LTP depends on the frequency (C) and number of bursts (D) of induction protocol. In (C), the burst number was fixed at 50 times, while the stimulation frequency was changed. In (D), the stimulation frequency was fixed at 100 Hz,
Neuron. Author manuscript; available in PMC 2017 September 22.
Hashimotodani et al.
Page 25
Author Manuscript Author Manuscript
while the burst number was changed. Numbers in parentheses, here and in all figures, indicate the number of cells. (E) MC-GC LTP was normally induced in the presence of 50 μM D-APV during the tetanus. (F) A pairing-protocol (200 pulses at 2 Hz, paired with 0 mV depolarization) also induced robust MC-GC LTP in the presence of 50 μM D-APV. Presynaptic activity alone (200 pulses, 2 Hz, Vh = −60 mV) also induced robust LTP. (G) In contrast, pairing-protocol-induced MPP LTP was abolished by D-APV, whereas presynaptic activity alone (200 pulses at 2 Hz) was not sufficient to induce LTP. (H) Recording configuration (top left), representative traces (bottom left), and summary plot time-course (right), showing that the induction protocol that elicits MC-GC LTP failed to induce LTP at DCG-IV-sensitive MPP inputs. (I) MC-GC LTP is input specific. Recording configuration (top left) and representative traces (bottom left). Summary plot (right) showing that MC-GC LTP was elicited at the tetanized, but not naive, input. Data are presented as mean ± SEM.
Author Manuscript Author Manuscript Neuron. Author manuscript; available in PMC 2017 September 22.
Hashimotodani et al.
Page 26
Author Manuscript Author Manuscript Author Manuscript
Figure 2. MC-GC LTP Is Likely Expressed Presynaptically
Author Manuscript
(A) MC-GC LTP was associated with a reduction of both PPR and CV. Representative traces are shown on left, summary plots (PPR and CV) on right (n = 18). (B) MC-GC LTP was similarly expressed at the AMPAR- and NMDAR-mediated components of synaptic transmission. Left: compound AMPAR/NMDAR EPSCs were recorded from GCs voltage clamped at −40 mV in the presence of 100 μM picrotoxin. Under control conditions, D-APV (50 μM) removed the slow component but had no effect on the peak amplitude of the compound EPSC. MC-GC LTP was assessed by simultaneously monitoring the AMPAR (peak measurement, red vertical line) and NMDAR component (offpeak measurement, blue vertical line) of MC EPSCs. Right: time-course summary plot showing that the AMPAR and NMDAR components exhibited similar LTP. (C) Representative experiment using minimal stimulation in IML; time course (bottom) and sample traces (top). (D) Summary plots demonstrating that MC-GC LTP induced by minimal stimulation is associated with a significant decrease in failure rate and increase in efficacy and potency (n = 8). (E) BoTx (5–500 nM) included in the intracellular solution did not affect MC-GC LTP.
Neuron. Author manuscript; available in PMC 2017 September 22.
Hashimotodani et al.
Page 27
Author Manuscript
(F) Control experiments confirming BoTx activity. Intracellulary applied BoTx (5–500 nM) through patch pipette induced rundown of AMPAR-EPSCs recorded from CA1 pyramidal neurons by stimulation of Schaffer collaterals. Data are presented as mean ± SEM. **p < 0.01, ***p < 0.001.
Author Manuscript Author Manuscript Author Manuscript Neuron. Author manuscript; available in PMC 2017 September 22.
Hashimotodani et al.
Page 28
Author Manuscript Author Manuscript Author Manuscript
Figure 3. MC-GC LTP Requires BDNF/TrkB Signaling, but Not CB1 Receptor Activity
Author Manuscript
(A) Representative traces (left) and time-course summary plot (right) showing that MC-GC LTP in the presence of the CB1 inverse agonist AM 251 (4 μM) was indistinguishable from MC-GC LTP control in interleaved slices. (B–D) MC-GC LTP was impaired in the presence of the TrkB inhibitor K252a (15 μM) (B), the TrkB receptor antagonist ANA-12 (15 μM) (C), or the BDNF scavenger TrkB-Fc (D) as compared to control human IgG (TrkB-Fc and IgG were used at 1 μg/mL, with a 20-min preincubation, and were also included in the perfusion). (E) Single-plane confocal images showing the dentate gyrus of TrkB-floxed mice injected with AAV5.CamKII.eGFP (left) or AAV5.CamKII.GFP-Cre (right). Note the presence of GFP-expressing cells in the granule cell layer of the dorsal blade and the absence of GFP expression in the hilus. (F) Representative traces (left) and time course (right) showing that MC-GC LTP was abolished in TrkB cKO GCs as compared to control GCs. Numbers in parentheses represent number of cells. Data are presented as mean ± SEM.
Neuron. Author manuscript; available in PMC 2017 September 22.
Hashimotodani et al.
Page 29
Author Manuscript Author Manuscript Author Manuscript Figure 4. BDNF Application Induces MC-GC LTP via a Presynaptic Mechanism
Author Manuscript
(A) BDNF induced LTP at MC-GC, but not MPP-GC, synapses. Representative traces (left) and time-course summary plot (right) showing that BDNF bath application (8 nM, 15 min) induced a long-lasting increase in the amplitude of MC-GC EPSCs, but not MPP-GC EPSCs, recorded in the same GC. (B) BDNF-induced MC-GC LTP was associated with a significant reduction of both PPR and CV. (C) Representative traces (left) and time-course summary plot (right) showing that brief puffs of BDNF in IML (vertical arrow) induced MC-GC LTP (white circles), and this
Neuron. Author manuscript; available in PMC 2017 September 22.
Hashimotodani et al.
Page 30
Author Manuscript
potentiation was abolished in the presence of the TrkB receptor antagonist ANA-12 (15 μM, gray circles). Puffing BDNF in the MML had no long-lasting effects on MPP-GC synaptic transmission. (D) BDNF-puff-induced MC-GC LTP was associated with a significant reduction of both PPR and CV. (E and F) BDNF puffs in IML increased sEPSC frequency but not amplitude. (E) Representative experiment showing traces (left) and sEPSC frequency versus time plot (right). Summary data are presented in (F). (G and H) No changes were observed by puffing BDNF in the presence of the TrkB antagonist ANA-12 (15 μM). Summary data are presented in (H). Numbers in parentheses represent number of cells. Data are presented as mean ± SEM. **p < 0.01, ***p < 0.001. n.s., not significant.
Author Manuscript Author Manuscript Author Manuscript Neuron. Author manuscript; available in PMC 2017 September 22.
Hashimotodani et al.
Page 31
Author Manuscript Author Manuscript Author Manuscript Figure 5. cAMP/PKA Signaling Is Involved in MC-GC LTP
Author Manuscript
(A) The PKA inhibitor H89 (10 μM, 40- to 60-min pre-incubation and also included in the perfusion) blocked MC-GC LTP compared to naive (control) hippocampal slices. (B) Loading PKI6–22 (2.5 μM) in GCs via recording pipette did not affect MC-GC LTP. (C) The effect of transient FSK bath application (50 μM) was assessed on MC and MPPmediated transmission in the same GC. FSK induced much larger potentiation at MC-GC synapses than at MPP synapses. Left: representative traces; middle: time-course summary plot of FSK-induced potentiation at MC- and MPP-GC synapses; right: PPR and CV of MC EPSCs before and after FSK application. (D) Preapplication of FSK (50 μM bath application for 10 min) occluded MC-GC LTP.
Neuron. Author manuscript; available in PMC 2017 September 22.
Hashimotodani et al.
Page 32
Author Manuscript
(E) After induction of MC-GC LTP, the stimulation strength was reduced to the original baseline (white arrow) to avoid a ceiling effect, and then, after obtaining a new baseline, FSK (50 μM, 10 min) was applied. (F) The cell-permeant PKA inhibitor PKI14–22 myristoylated (1 μM bath application) blocked BDNF-induced MC-GC LTP compared to interleaved controls. (G) FSK-induced LTP (50 μM, 10 min) recorded from cKO and control GCs. Numbers in parentheses represent number of cells. Data are presented as mean ± SEM. *p < 0.05, ***p < 0.001.
Author Manuscript Author Manuscript Author Manuscript Neuron. Author manuscript; available in PMC 2017 September 22.
Hashimotodani et al.
Page 33
Author Manuscript Author Manuscript Author Manuscript Author Manuscript
Figure 6. Optically Elicited MC EPSCs Show FSK-LTP and MC-GC LTP
(A) Schematic diagram showing the injection of the AAV-hSyn-ChR2(H134R)-EYFP or AAV-hSyn-ChIEF-citrine in the hilus. (B–E) Infrared/differential interference contrast (IR/DIC) (B and D) and fluorescence images (C and E) showing that ChR2(H134R)-EYFP was selectively expressed in the IML of contralateral dentate gyrus. (F) Schematic diagram showing optic stimulation of ChR2(H134R)-EYFP-expressing MC axons in contralateral hippocampal slices.
Neuron. Author manuscript; available in PMC 2017 September 22.
Hashimotodani et al.
Page 34
Author Manuscript Author Manuscript
(G) Optically evoked EPSCs (O-EPSCs) recorded from GCs were blocked by TTX (0.5 μM) (top) or sequential application of NBQX (10 μM) and D-APV (50 μM) (bottom). (H) Representative traces (left) and time-course summary plot (right) showing that, as expected for MC inputs, but not MPP inputs, O-EPSCs were insensitive to bath application of DCG-IV (1 μM). (I) Top: example of optically induced synaptic responses during the LTP protocol (5 pulses, 30 Hz, repeated 50 times at 2 Hz) in hippocampal slices expressing ChIEF-citrine. Insets: first three and last three burst-induced responses are shown on an expanded timescale. Bottom, representative traces (left) and time-course summary plot (middle) showing robust optically induced LTP, which was accompanied by a significant reduction in PPR and CV (right). (J) Representative traces (left) and time-course summary plot (middle) showing that transient bath application of FSK (50 μM, 10 min) induced long-lasting potentiation of OEPSCs, and this potentiation was accompanied by a significant reduction of PPR and CV (right). Data are presented as mean ± SEM. **p < 0.01, ***p < 0.001.
Author Manuscript Author Manuscript Neuron. Author manuscript; available in PMC 2017 September 22.
Hashimotodani et al.
Page 35
Author Manuscript Author Manuscript
Figure 7. Assessing Excitatory and Inhibitory Transmission and Plasticity by Optogenetic Activation of MC Axons
Author Manuscript Author Manuscript
(A) Left: schematic diagram illustrating ChR2(H134R)-EYFP infected MCs project to the contralateral GCs and hilar interneurons. Middle: sample traces showing O-EPSC-IPSC sequence recorded from GCs voltage clamped at −60 mV. Note that ECl− is approximately −89 μV. Biphasic currents were abolished by 10 μM NBQX and 50 μM D-APV (middle, red trace). Summary data are shown on the right. (B) Sample traces showing O-dIPSC was blocked by 100 μM picrotoxin (blue trace) and then O-EPSC was abolished by subsequent application of 10 μM NBQX and 50 μM D-APV (red trace). Right: summary data showing the amplitude of O-dIPSCs (at time point 30 ms after light illumination) before and after application of picrotoxin (100 μM). (C) Scatterplot of O-EPSC and O-IPSC amplitudes. Each amplitude was analyzed at peak currents, which underestimates real amplitudes of EPSCs/dIPSCs. Black square shows the mean value. Note that dotted line denotes unity. (D) Simultaneous recordings of O-EPSCs and O-dIPSCs before and after LTP induction. Representative traces (left) and summary time course (right) showing that optically delivered LTP induction (same as Figure 6I) induced LTP at O-EPSCs, but not at O-dIPSCs. To confirm the disynaptic nature of O-IPSCs, 10 μM NBQX and 50 μM D-APV were applied at the end of every experiment (red trace). The amplitudes of O-EPSCs were measured at peak inward currents and O-dIPSCs were measured at peak outward currents. Dotted line shown in traces indicates time point corresponding to 30 ms after light illumination. (E) Left top: schematic diagram illustrating ChIEF-citrine-infected MCs project to the contralateral GCs and hilar interneurons. Left bottom: z stack two-photon fluorescence images showing an example of Alexa-Fluor-594-filled interneuron in the molecular layer (ML). Right: time-course summary plot showing that repetitive MC axon photostimulation (5 pulses at 30 Hz, repeated 50 times at 2 Hz) did not induce long-lasting synaptic plasticity. Inset: representative O-EPSCs before and after repetitive stimulation. O-EPSCs were recorded in the continuous presence of 100 μM picrotoxin. Data are presented as mean ± SEM.
Neuron. Author manuscript; available in PMC 2017 September 22.
Hashimotodani et al.
Page 36
Author Manuscript Author Manuscript Author Manuscript
Figure 8. Functional Consequences of MC-GC LTP on the GC Output
Author Manuscript
(A and B) Representative experiment showing GC firing elicited by burst stimulation in IML (5 pulses, 20 Hz) before and after bath application of picrotoxin (100 μM). (A) Sample traces. (B) Time-course plot of the number of spikes per burst. (C) Summary data showing the spike probability before and after application of picrotoxin. (D and E) Representative experiment showing GC firing elicited by burst stimulation in IML (5 pulses, 20 Hz), before and after delivering the LTP induction protocol. (D) Sample traces. (E) Time-course plot of the number of spikes per burst showing the long-lasting enhancement of MC-driven GC firing after LTP induction. (F and G) Sample traces of two representative experiments showing burst stimulation before and after application of the LTP induction protocol in the presence of 15 μM ANA-12 (F) or 10 μM H89 (G). (H) Summary plot showing the spike probability under three different conditions. The increase in spike probability after application of the protocol was abolished either in the presence of 10 μM H89 or 15 μM ANA-12. Data are presented as mean ± SEM. *p < 0.05, **p < 0.01. n.s., not significant.
Neuron. Author manuscript; available in PMC 2017 September 22.