Synaptic plasticity is a process whereby synapses can be strengthened or weakened by either presynaptically altering neurotransmitter release, or postsynaptically modifying synaptic receptor numbers. Plasticity is a critical attribute that allows for brain modification in an experience-dependent fashion, with the hippocampus encoding and consolidating memory. Two major synaptic plasticity forms exhibited in the brain are long-term potentiation (LTP) [1
] and long-term depression (LTD) [2
]. Hippocampal LTP of cornu ammonis (CA1) pyramidal cells is dependent on N
-aspartate (NMDA) glutamate receptors, which induce LTP by triggering a signal cascade that results in the insertion of α-amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid (AMPA) glutamate receptors into the postsynaptic membrane [3
]. Other NMDA receptor-independent forms of plasticity also occur that are mediated by various mechanisms including lipid messengers known as endocannabinoids (eCBs).
Endocannabinoids are normally produced by catabolism of phosphatidylinositol 4,5-bisphophate from the lipid membrane. The various enzymes involved in the process include n-acylphosphatidylethanolamine phospholipase D (NAPE-PLD), diacylglycerol lipase alpha (DAGLα), and 12-lipoxygenase, which normally follow G-protein second messenger activation [4
]. For example, activation of postsynaptic type I metabotropic glutamate receptors can result in formation of metabolites that are modified to produce eCBs or eicosanoid. These lipid molecules often act retrogradely on presynaptic receptors such as cannabinoid receptor 1 (CB1) to alter neurotransmission [9
]. The two best-characterized and most widely expressed endocannabinoids are anandamide and 2-arachadonyl glycerol (2-AG). Anandamide is produced by several pathways, including by NAPE-PLD, and can bind to CB1, transient receptor potential vanilloid 1 (TRPV1) [6
], and other G-protein-coupled receptors [11
]. Anandamide is degraded by the enzyme fatty acid amide hydrolase (FAAH) [7
]. 2-AG is produced in the brain by DAGLα and involved in retrograde signaling via CB1 [8
]. Finally, the enzyme 12-lipoxygenase can synthesize 12-(S
-tetraenoic acid (12-HPETE), which can activate TRPV1 receptors [13
The hippocampus contains both excitatory and inhibitory neurons. Excitatory pyramidal cells are located in the stratum pyramidale layer and diverse inhibitory interneurons are present in layers such as the stratum radiatum and stratum oriens [14
]. Hippocampal interneurons are a diverse group, and can be classified into several subtypes based on their function, axon projections, and firing pattern. These subtypes are often characterized by expression of calcium-binding proteins and neuropeptides, including somatostatin (SOM) and parvalbumin (PV). Hippocampal CA1 stratum oriens interneuron subtypes include oriens lacunosum-moleculare (O-LM) interneurons, which can be identified by the expression of somatostatin and have regular-to-fast action potential spiking patterns [15
]. O-LM cell soma and dendrites reside in the stratum oriens and their axons project to the stratum lacunosum-moleculare layer. Another subtype is parvalbumin-positive basket cells, which project locally [18
]. These basket cells are also involved in network oscillations in the hippocampus [19
Stratum oriens interneurons exhibit plasticity, which is markedly unique compared to neighboring layers such as the stratum radiatum. For example, the type I mGluR (mGluR1 and mGluR5) agonist (S
)-3,5-DHPG will depress glutamatergic synaptic responses onto radiatum interneurons [20
], but potentiate glutamatergic synaptic responses onto oriens interneurons [21
]. LTP is induced in O-LM neurons by high-frequency stimulus (HFS; 100 Hz) and is anti-Hebbian due to the requirement for postsynaptic hyperpolarization, which is needed to relieve a polyamine block of calcium-permeable AMPA receptors [22
]. While this LTP is induced postsynaptically and depends on mGluRs, post-synaptic calcium, M1 muscarinic receptors, and alpha7 nicotinic acetylcholine receptors [23
], it is expressed presynaptically. However, the full presynaptic mechanism has yet to be fully characterized and any potential involvement of eCBs in this plasticity is unknown, though it is independent of nitric oxide and TRPV1 presynaptic signaling [16
]. Our goal was to determine whether interneuron LTP is dependent on presynaptic eCB signaling and whether interneurons have the machinery to produce these eCBs. While it is most common for eCB plasticity via CB1 receptors to induce LTD, CB1-dependent LTP was recently observed in both the hippocampal lateral perforant pathway [26
] and in cortical inputs onto dentate granule cells [27
]. Collectively, based on the role oriens interneurons play in feedback inhibition of CA1 pyramidal cells, and widespread network control including oscillations, and given that they undergo plasticity, it is important to fully characterize this novel form of LTP. Here, we report that LTP in stratum oriens interneurons is CB1-dependent and that oriens interneurons express eCB biosynthetic enzymes.
The goal of our study was to investigate the role of endocannabinoid signaling in stratum oriens interneuron LTP. Therefore, we employed whole-cell patch clamp electrophysiology to identify plasticity of stratum oriens interneurons followed by real-time PCR (RT-PCR) on the extracted cell to examine mRNA components involved in endocannabinoid biosynthesis. To induce LTP, we applied a 100 Hz HFS to the stratum oriens layer near the recorded cell. LTP was evoked in 67% of interneurons (Figure 1
A); the remaining cells evoked glutamate responses, but did not exhibit plasticity. To examine the involvement of eCBs and CB1 in this LTP, we applied the CB1 inverse agonist AM-251 to the slices for at least 15 min before LTP induction and observed LTP was blocked in all cases (Figure 1
B). Controls with no HFS resulted in neither significant potentiation nor depression, and no change in EPSCs (data not shown; n
= 6, p
> 0.5). After the recording, cells were extracted in order to examine cellular mRNA. Classification of cells as interneurons was based on GAD65 and/or GAD67 mRNA expression and their location in the stratum oriens away from stratum pyramidale. In this way, interneurons were classified into four subtypes based on their expression of either somatostatin (SOM+), parvalbumin (PV+), calbindin (CB+), or calretinin (CR+) (see Table 1
and see examples in Figure 2
). We further observed that in general, SOM+ cells were those exhibiting LTP in Figure 1
A), while SOM-negative cells, such as PV+ cells, did not (Figure 2
B). Furthermore, LTP of SOM+ neurons was blocked in the presence of AM-251 (Figure 2
C). Important to note here is that while the majority of LTP-exhibiting cells were SOM+ cells, one LTP cell was PV+, SOM-negative. However, we cannot determine whether this LTP cell was potentially a SOM-false negative cell, as some SOM+ cells also express PV [28
], or a true PV+ basket cell. In addition, another LTP cell was included in the physiology though it was non-classifiable, likely due to false negatives for interneuron marker subtypes. One non-LTP cell was also unclassified. Collectively, the data demonstrate that CB1-dependent LTP mainly occurs in SOM+ cells.
Because CB1 receptors are normally involved in synaptic depression rather than potentiation, we wanted to confirm CB1 involvement in synaptic enhancements using a different approach. To do this, we recorded glutamate EPSCs from oriens interneurons before and after a fatty acid amide hydrolase (FAAH) inhibitor, URB597, was applied to the extracellular bath solution. FAAH inhibition prevents anandamide hydrolysis and increases synaptic levels of anandamide. We noted that FAAH inhibition produced a significant (p
< 0.05) potentiation compared to baseline (Figure 3
A). This potentiation was blocked by the CB1 antagonist AM-251 (Figure 3
B), demonstrating CB1 dependence of anandamide-mediated synaptic enhancements. In addition, we applied the monoacylglycerol lipase (MAG lipase) inhibitor, JZL184 (1 µM), which prevents degradation of 2-AG. This resulted in significant (p
< 0.05) enhancement of EPSC responses (Figure 3
C) as well, supporting the FAAH inhibitor data and the involvement of endocannabinoids in potentiation of these excitatory inputs.
To examine presynaptic location of LTP and FAAH-induced potentiation we performed paired pulse ratios as described previously [30
] and noted both a trend for depressed ratios after HFS or URB597 application (data not shown; both had p
< 0.1; see Figure 1
A LTP traces). This suggested LTP is potentially presynaptic, which has been illustrated by many others examining oriens SOM+ interneuron LTP as described previously; thus, our data, while only trending, support that of others.
As several groups, including ours, have data supporting the ability for various GABAergic interneurons, including those in the stratum radiatum, to produce eCBs and thus potentially modify their own activity, we examined this question using RT-PCR and immunohistochemistry (IHC). RT-PCR revealed that the most abundant interneuron, the SOM+ cells, expressed mRNA for DAGLα, NAPE-PLD, 12-lipoxygenase, and/or type I mGluRs (see Table 1
). PV+ cells were the second-largest group, and expressed mRNA for NAPE-PLD, 12-lipoxygenase, and/or type I mGluRs, while CB+ cells had only one such cell expressing eCB enzyme mRNA. CR+ neurons were the minority, and did not express any of the eCB-related proteins we tested. This suggests stratum oriens interneurons, particularly SOM+ and PV+ cells, but not CB/CR cells, have the potential to produce eCBs. While we performed physiology on 33 cells where full PCR was carried out, not all were classifiable and were thus excluded from the RT-PCR expression data (n
= 29 total, all of which expressed control 18S at a reliable level to confirm successful mRNA harvesting; Table 1
). Lastly, while potential false negatives likely led to low expression levels of both eCB-producing enzymes and type I mGluRs (see discussion for further explanation), expression of DAGLα in some cells that exhibited LTP was noted. However, this alone cannot confirm interneuron DAGLα is used for CB1-dependent LTP, but alternatively could also be used for another purpose.
Lastly, to confirm that mRNA was actually translated to protein we employed IHC to examine protein expression of DAGLα, the enzyme producing the eCB 2-AG, within oriens interneurons of rodents where green fluorescent protein (GFP) was used as an interneuron marker. Mice, rather than rats, were used in this case as we employed a modified mouse line where GAD67-positive cells also express GFP, which was used in place of notoriously poor GAD antibodies, allowing us to make a positive identification of interneurons genetically. These GAD67-postive cells were demonstrated previously to include SOM+ and PV+ cells [31
]. DAGLα exhibited cytosolic expression in 64% of GAD67-positive cells examined (Figure 4
). Expression of DAGLα was noted in interneurons of the stratum radiatum as well. Collectively, our data indicate that SOM+ stratum oriens LTP is CB1-dependent, eCBs potentiate excitatory transmission to oriens interneurons, and SOM+ and PV+ interneurons have at least the potential to produce eCBs based on molecular studies.
All experiments were performed in accordance with Institutional Animal Care and Use Committee protocols and followed the NIH guidelines for the care and use of laboratory animals. Male CD1 GAD67-GFP mice were used in immunohistochemical experiments. Male Sprague–Dawley rats age 15–34 days old were used for all electrophysiological experiments.
Rats were anesthetized with isoflurane and decapitated. Brains were removed and sectioned coronally on a vibratome at 400 µm. Recordings began at least one hour after cutting while tissue was stored in oxygenated artificial cerebral spinal fluid (ACSF) composed of 119 mM NaCl, 26 mM NaHCO3, 2.5 mM KCl, 1 mM NaH2PO4, 2.5 mM CaCl2, 1.3 mM MgSO4, and 11 mM glucose.
Whole cell voltage-clamp experimental methods similar to these were described previously [10
]. Briefly, hippocampal CA1 oriens cells were visualized using an Olympus BX51WI microscope (Center Valley, PA, USA) with a 40× water immersion objective. Cells were patched with a glass pipette filled with internal solution composed of 117 mM cesium gluconate, 2.8 mM NaCl, 20 mM HEPES, 5 mM MgCl2
, and QX-314 (Tocris) (pH 7.28, 275–285 mOsm). Polyamines were omitted from the internal solution as done previously by others examining oriens interneuron LTP [16
] so that calcium permeable AMPA receptors known to be required for LTP were not under voltage-dependent block. Picrotoxin (100 µM; Abcam) was added to the ACSF recording solution to block GABAA
receptor currents. AM-251, JZL184 and URB597 were purchased from Tocris (Bristol, UK). Cells were recorded in voltage-clamp at −65 mV. Stimulation to induce evoked synaptic transmitter release was accomplished using a concentric bipolar stimulating electrode placed in the stratum oriens layer near the cell recorded from to induce activation of glutamatergic axons including CA1 recurrent collaterals and potentially Schaffer Collateral inputs, along with other local neurotransmitter inputs. Stimulation ranged from 60 to 300 µA and evoked paired responses separated by 50 msec. Baseline and post-HFS stimulation was at 0.1 Hz for a 100 µsec duration to evoke EPSCs and the cell was recorded from for as long as possible. Baseline traces were acquired for approximately 10 min with no experiment going beyond ~12 min post-whole cell acquisition in order to avoid potential washout. Plasticity was induced using 2 stimulations at 100 Hz for 1 s (100 µsec stimulus duration), 20 s apart, while the cell was in current clamp mode, then recording resumed in voltage clamp mode. For all whole cell experiments, traces were recorded using Multiclamp 700B amplifier (Molecular Devices, Sunnyvale, CA, USA). Signals were filtered at 4 kHz and digitized with an Axon 1440A or 1550A digitizer (Molecular Devices) connected to a Dell personal computer with pClamp 10.2 or 10.5 Clampfit software (Molecular Devices). p
-Values were obtained using a two-way unequal variance student’s t
-test or ANOVA with p
< 0.05 being considered significant, taken 10–15 min post-conditioning or drug application.
4.2. Reverse Transcription and Pre-Amplification
RT-PCR methods were carried out similar to prior methods [10
]. Briefly, cells used for RT-PCR analysis were extracted using gentle suction and placed into chilled reverse transcriptase reagents (BioRad, Hercules, CA, USA) and processed within 2 h. The entire cell was harvested to increase the amount of mRNA and thus reduce false negatives and attain as much starting mRNA as possible, as we had a large number of targets to examine. As a result, filling and post-hoc tracing of cells was untenable. One control sample of artificial cerebral spinal fluid was obtained for each slice and used to identify false positives due to contamination from extracellular mRNA, as detailed previously [29
]. Using iScript cDNA Synthesis kit (BioRad), extracted cells were reverse transcribed to cDNA under the manufacturer’s protocol and cycled in a C1000 Thermocycler (BioRad) at 25 °C for 8 min, 42 °C for 60 min, and 70 °C for 15 min. Following reverse transcription, each cell was divided into three 5 µL aliquots, each of which received a different group of 10-fold diluted primers, iQ Supermix (BioRad), and ddH2
O. The samples were then cycled in a C1000 Thermocycler (BioRad) starting at 95 °C for 3 min, then 15 cycles of 95 °C for 15 s, 57 °C for 20 s, and 72 °C for 25 s. Additionally, no template control multiplex tests were done to ensure there were no primer dimer or hairpin interactions among grouped multiplex primer sets that would interfere with later individual runs.
4.3. Primer Design
Most primer and probe sequences were used in a prior study [10
]. Primers were designed for somatostatin using the same parameters, efficiency, and melting temperature with Vector NTI software. The somatostatin forward primer, reverse primer, and probe sequences used were ACCCCAGACTCCGTCAGTTTC, GTTGGGCTCAGACAGCAGTTCT, and ACCGGGAAACAGGAACTGGCCAAGT, respectively. The appropriate fluorescent 6-carboxyfluorescein-tetramethylrhodamine (FAM-TAMRA) probe (Applied BioSystems, Inc., Waltham, MA, USA) for each target was included in the RT-PCR reaction to ensure specificity, as each individual probe was designed to bind only to its unique target amplicon, thereby eliminating alternative non-specific binding products of the primers from fluorescent analysis. Validation of these primers (gels, etc.) were illustrated in two prior publications [10
]. Therefore, our RT-PCR data using specific probe-based fluorescence is more accurate in analysis than gels (which can still demonstrate non-specific products). As a positive reference/housekeeping control, we examined 18S to ensure successful harvesting of the cells. 18S values ranged from 14–18 on their cycle threshold (Ct) after multiplexing, similar to that reported by our lab previously [10
]. If 18S was greater than 20, it was excluded from our analysis. Therefore, 18S PCR demonstrates successful completion of harvesting of mRNA, reverse transcription reaction, and multiplexing of these cells. In summary, 18S was used simply to confirm RT-PCR results of all other targets, but was not used to determine ΔCt values used to quantify differences in RT-PCR of these targets between cells, which we assume not to be significantly different from each other based on our prior study of stratum radiatum interneurons [10
]. We illustrate raw RT-PCR Ct fluorescent curves from FAM-TAMRA probes to confirm expression of mRNA of these targets. Important to note is that we attempted to quantify expression of CB1 receptors as well, however as CB1 is the most highly expressed G-protein-coupled receptor in the brain it always came up in our ACSF controls as background, and therefore was eliminated from our analysis of single cells as a potential false positive.
4.4. Quantitative RT-PCR Reaction
Each pre-amplified cell was run for every target individually and in triplicate with its appropriate primer set and probe. Each cell was run in a CFX96 RT-PCR machine (BioRad) with a 95 °C hot start for 3 min, followed by 60 cycles of 95 °C for 15 s, 57 °C for 25 s, and 72 °C for 25 s. Cycle threshold values were determined with BioRad CFX manager 3.1 software.
Male GAD67-GFP knock-in mice (20–30 days old) were used for immunohistochemical experiments. Animals were anesthetized with isoflurane and perfusion fixated (cardiac) with 0.9% NaCl, then 4% paraformaldehyde. The brains were removed and placed in paraformaldehyde where they remained overnight. They were then cryoprotected in 30% sucrose before being sectioned coronally through the hippocampus at 30 μm using a Microm HM 550 cryostat (Richard-Allan Scientific, Kalamazoo, MI). Free-floating sections were stored in 1 M PBS (pH 7.4), then washed in 0.2% Triton X in PBS for 30 min, a blocking solution of 1% BSA, 5% normal goat serum in PBS for 2 h, then in primary antibody overnight. The following day the sections were again washed in Triton X blocking solution, followed by the secondary antibody for 2 h. Slices were washed in 1 M tris buffered saline, and then mounted on non-frosted glass slides (Fisher Scientific, Pittsburgh, PA, USA) and cover-slipped using Fluoromount G (Southern Biotech, Birmingham, AL, USA). Images were taken using Olympus FluoView FV1000 confocal laser scanning microscope at 20× (Center Valley, PA, USA). The concentration of antibody used was DAGLα 1:500 (kindly provided by Dr. Ken Mackie), and AlexaFluor 350 goat anti-rabbit secondary 1:1000 (Invitrogen). An attempt to confirm NAPE-PLD expression by IHC was unsuccessful as we lacked confidence in the specificity of the antibodies we employed.