Expression of TRPV1 channels by Cajal‐Retzius cells and layer‐specific modulation of synaptic transmission by capsaicin in the mouse hippocampus

Abstract

Key points

• By taking advantage of calcium imaging and electrophysiology, we provide direct pharmacological evidence for the functional expression of TRPV1 channels in hippocampal Cajal‐Retzius cells.

• Application of the TRPV1 activator capsaicin powerfully enhances spontaneous synaptic transmission in the hippocampal layers that are innervated by the axons of Cajal‐Retzius cells.

• Capsaicin‐triggered calcium responses and membrane currents in Cajal‐Retzius cells, as well as layer‐specific modulation of spontaneous synaptic transmission, are absent when the drug is applied to slices prepared from TRPV1⁻/⁻ animals.

• We discuss the implications of the functional expression of TRPV1 channels in Cajal‐Retzius cells and of the observed TRPV1‐dependent layer‐specific modulation of synaptic transmission for physiological and pathological network processing.

Abstract

The vanilloid receptor TRPV1 forms complex polymodal channels that are expressed by sensory neurons and play a critical role in nociception. Their distribution pattern and functions in cortical circuits are, however, much less understood. Although TRPV1 reporter mice have suggested that, in the hippocampus, TRPV1 is predominantly expressed by Cajal‐Retzius cells (CRs), direct functional evidence is missing. As CRs powerfully excite GABAergic interneurons of the molecular layers, TRPV1 could play important roles in the regulation of layer‐specific processing. Here, we have taken advantage of calcium imaging with the genetically encoded indicator GCaMP6s and patch‐clamp techniques to study the responses of hippocampal CRs to the activation of TRPV1 by capsaicin, and have compared the effect of TRPV1 stimulation on synaptic transmission in layers innervated or non‐innervated by CRs. Capsaicin induced both calcium responses and membrane currents in ∼50% of the cell tested. Neither increases of intracellular calcium nor whole‐cell currents were observed in the presence of the TRPV1 antagonists capsazepine/Ruthenium Red or in slices prepared from TRPV1 knockout mice. We also report a powerful TRPV1‐dependent enhancement of spontaneous synaptic transmission onto interneurons with dendritic trees confined to the layers innervated by CRs. In conclusion, our work establishes that functional TRPV1 is expressed by a significant fraction of CRs and we propose that TRPV1 activity may regulate layer‐specific synaptic transmission in the hippocampus. Lastly, as CR density decreases during postnatal development, we also propose that functional TRPV1 receptors may be related to mechanisms involved in CR progressive reduction by calcium‐dependent toxicity/apoptosis.

Key points

• By taking advantage of calcium imaging and electrophysiology, we provide direct pharmacological evidence for the functional expression of TRPV1 channels in hippocampal Cajal‐Retzius cells.

• Application of the TRPV1 activator capsaicin powerfully enhances spontaneous synaptic transmission in the hippocampal layers that are innervated by the axons of Cajal‐Retzius cells.

• Capsaicin‐triggered calcium responses and membrane currents in Cajal‐Retzius cells, as well as layer‐specific modulation of spontaneous synaptic transmission, are absent when the drug is applied to slices prepared from TRPV1⁻/⁻ animals.

• We discuss the implications of the functional expression of TRPV1 channels in Cajal‐Retzius cells and of the observed TRPV1‐dependent layer‐specific modulation of synaptic transmission for physiological and pathological network processing.

Introduction

CRs are early‐generated neurons that play important developmental roles in orchestrating the building of the hippocampal and neocortical architecture (Gil et al. 2014). In contrast to the neocortex, where CRs undergo rapid bax/caspase‐3‐dependent apoptosis and virtually disappear after the first 2–3 postnatal weeks (Chowdhury et al. 2010; Anstötz et al. 2014; Ledonne et al. 2016), CRs remain in the hippocampus until maturity, albeit at lower densities (Supèr et al. 1998; Anstötz et al. 2016, 2018). Density reduction is obtained by caspase‐3 independent cell death (Anstötz et al. 2016; Ledonne et al. 2016), but the underlying molecular mechanisms and signalling pathways remain currently unknown. CRs innervate the molecular layers of the dentate gyrus, stratum lacunosum‐moleculare of the Cornu Ammonis, and the interposed hippocampal fissure (von Haebler et al. 1993; Marchionni et al. 2010; Quattrocolo & Maccaferri, 2014; Anstötz et al. 2016, 2018). Here, synaptic terminals of CRs release glutamate onto GABAergic interneurons and pyramidal cells, thus activating AMPA‐ and NMDA‐type postsynaptic receptors (Quattrocolo & Maccaferri, 2014; Anstötz et al. 2016, 2018). Because of these morpho‐functional properties, CRs have been proposed to participate in hippocampal layer‐specific signal integration (Maccaferri, 2011).

Despite the fact that CRs form an excitatory loop in addition to the classical ‘trisynaptic circuit’ (Andersen, 1975), their physiological (or pathological) role in the regulation of hippocampal activity in vivo is currently poorly understood.

Cavanaugh et al. (2011) have recently suggested that hippocampal CRs are the only cell type in the hippocampus expressing TRPV1 (for alternative views, please see below and Results and Discussion sections). These receptors form non‐selective cation channels that can be activated by a variety of stimuli/agonists such as temperature, protons, vanilloids like capsaicin, and some membrane‐derived lipids (Caterina et al. 1997; see review by Rosenbaum & Simon, 2007). Although many studies have explored the role of TRPV1 in mediating pain and inflammatory signals from peripheral neurons (Tominaga et al. 1998; see reviews by Caterina & Julius, 2001 and Julius, 2013, accumulating evidence suggests that TRPV1 play also important roles in the central nervous system such as regulating membrane potential, neurotransmitter release and synaptic plasticity (see review by Kauer & Gibson, 2009). In the hippocampus, the expression of TRPV1 in CRs was suggested by two TRPV1 reporter mice (Cavanaugh et al. 2011). However, the same study could not detect TRPV1‐dependent calcium responses in intact slices of the dentate gyrus (which is a hippocampal region containing CRs, see von Haebler et al. 1993; Marchionni et al. 2010; Anstötz et al. 2014, 2016). Therefore, both the functionality of TRPV1 expressed by CRs and its potential role in the regulation of the hippocampal network remain undetermined.

These are, nevertheless, important points. In fact, activation of TRPV1‐mediated signalling in CRs might activate calcium‐dependent intracellular pathways, depolarize their membrane potential and/or alter transmitter release. This latter action might modulate synaptic transmission onto postsynaptic GABAergic interneurons and pyramidal cells, eventually leading to modified hippocampal rhythms, synchronization and overall excitability.

Here, we have begun to clarify these issues by directly recording TRPV1‐mediated calcium responses and currents in mouse hippocampal CRs, and by studying the effect of TRPV1 activation on layer‐specific hippocampal glutamatergic synaptic transmission.

Methods

Ethical approval

All procedures used in this study were in compliance with the guidelines provided by the Institutional Animal Care and Use Committee of Northwestern University (approved protocol: IS00004871) and set forth in the NIH Guide for the Care and Use of Laboratory Animals. The investigators understand the ethical principles under which The Journal of Physiology operates and our work complies with its animal ethics checklist.

Animals

Both male and female mice were used. For calcium imaging experiments, animals were generated by breeding Wnt3a‐IRES‐Cre mice (strain: B6(Cg)‐Wnt3a^{tm1.1(cre)Mull}/Mmmh; stock number: 031748‐MU; RRID: IMSR_MMRRC:031748; see Gil‐Sanz et al. 2013) with animals that conditionally express the calcium indicator GCaMP6s (Chen et al. 2013) following Cre‐mediated removal of the floxed STOP cassette (from Jackson Laboratory, Bar Harbor, ME, USA; strain: B6;129S6‐Gt(ROSA)26Sor^{tm96(CAG‐GCaMP6s)Hze}/J; stock number: 024106; RRID: IMSR_JAX:024106; henceforth referred to as Wnt3a‐GCaMP6s mice).

For electrophysiological experiments, we used either wild‐type C57BL/6J animals (stock number: 000664; RRID: IMSR_JAX:000664) or the following reporter mice that facilitate the identification of CRs: CXCR4‐EGFP (strain: STOCK Tg(Cxcr4‐EGFP)CD73Gsat/Mmucd (MMRRC ID: 015859‐UCD); RRID: MMRRC_015859‐UCD; see: Marchionni et al. 2010), or the animals generated by crossing Wnt3a‐IRES‐Cre mice with a floxed tdTomato reporter (strain: B6.Cg‐Gt(ROSA)26Sor^{tm9(CAG‐tdTomato)Hze}/J; stock number: 007909; RRID: IMSR_JAX:007909; see Anstötz et al. 2018). For experiments requiring the functional elimination of TRPV1 responses appropriate breeding with TRPV1 KO mice (strain: B6.129 × 1‐Trpv1^tm1Jul/J; stock number: 003770; RRID: JAX:003770) was performed to obtain CXCR4‐EGFP TRPV1⁻/⁻ mice.

Immunocytochemistry

Animals were anaesthetized by intraperitoneal injection of sodium pentobarbital and perfused with isotonic saline followed by 4% paraformaldehyde in 0.12 M phosphate buffer (PB), pH 7.4. After perfusion, brains were extracted from the skulls, and cryoprotected in 30% sucrose in PBS. Horizontal hippocampal sections were cut serially at 60 μm on a freezing‐stage microtome, then incubated in TBS containing 5% NGS (normal goat serum), 1% BSA (bovine serum albumin) and 0.2% Triton X‐100 for 1 h at room temperature. After carrying out the specific immunocytochemical reactions followed by wash in TBS, sections were coverslipped with Mowiol and observed under a confocal microscope.

GFP

Primary anti‐GFP antibody (1:1000, rabbit; catalogue no. G10362 (Thermo Fisher Scientific, Waltham, MA, USA); RRID) was applied in TBS containing 1% NGS, 1% BSA and 0.2% Triton X‐100 at 4°C overnight. The secondary antibody used was conjugated to Alexa488 (1:500, goat; catalogue no. A11034 (Thermo Fisher Scientific); RRID: AB_2576217). This reaction was used either to identify GCaMP6s in Wnt3a‐GCaMP6s mice (P11, n = 3, and P12, n = 2), or EGFP in CXCR4‐EGFP TRPV1⁺/⁺ (n = 6, P30) and CXCR4‐EGFP TRPV1⁻/⁻ animals (n = 6, P30).

TdTomato

Sections were incubated with a primary polyclonal antibody recognizing tdTomato (1:3000, rabbit; catalogue no. 632496 (Takara Bio USA, Inc., Mountan View, CA, USA); RRID: AB_10013483) and a following secondary antibody conjugated to Alexa594 (1:500, goat; catalogue no. 11037 (Thermo Fisher Scientific); RRID: AB_2534095). Three Wnt3a‐tdTomato mice (P30) were used.

Anatomical recovery of biocytin‐filled cells

Slices containing cells filled with biocytin during whole‐cell recording were fixed overnight in 4% paraformaldehyde in 0.1 M PB at 4 C. Endogenous peroxidase activity was quenched with 3% H₂O₂ solution for 15 min. Sections were incubated overnight at 4 C in avidin‐biotinylated‐HRP complex (Vectastatin ABC Elite Kit, Vector Laboratories, Burlington, ON, Canada) with 0.1% Triton X‐100 in 0.1 M PB, followed by a peroxidase reaction with 3,3′‐diaminobenzidine tetrahydrochloride as a chromogen, and then intensified with 1% NiNH₄SO₄ and 1% CoCl_2. Cells were checked for contrast under light microscopy and briefly postfixed with 0.1% OsO₄ in 0.1 M PB for 1–2 min, mounted on slides, and coverslipped with Mowiol. Cells were reconstructed using a NEUROLUCIDA‐based station and software (MBF Bioscience, Williston, VT, USA).

Slice preparation for calcium and electrophysiological recordings

Animals in the P12–P21 age range were used. After deep anesthesia with isoflurane and decapitation, the brain of the animal was gently extracted and glued to a small container filled with chilled modified artificial cerebrospinal fluid (ACSF) of the following composition (in mM): 130 NaCl, 24 NaHCO₃, 3.5 KCl, 1.25 NaH₂PO₄, 1 CaCl₂, 2 MgSO₄, 10 glucose, saturated with 95% O₂, 5% CO₂ at pH = 7.4. Transverse sections were cut (300 μm and 400 μm thick for calcium imaging and electrophysiological recordings, respectively) using a vibrating microtome (Leica VT 1200 S, Leica Biosystems, Heidelberg, Germany). After preparation, slices were incubated at 31–33˚C for at least 30 min and then stored at room temperature.

Calcium recordings and analysis

Calcium imaging was performed from tissue obtained from Wnt3a‐GCaMP6s mice. Slices were imaged using an epifluorescence microscope equipped with a 60× objective (NA 1.0) and a collimated LED (460 nm, Prizmatix, Holon, Israel) used as an excitation source coupled to a GFP filter set (excitation band: 469 ± 17.5 nm; emission band: 525 ± 19.5 nm; dichroic reflection/transmission band 452–490 nM/505–800 nm; Thor labs). LED‐generated flashes (40 ms duration) at 4 Hz were synchronized with the acquisition of emitted fluorescence by a Zyla sCMOS 4.2 camera (Andor, Springvale Business Park, Belfast, UK).

Raw images were imported and analysed using the μManager (Edelstein et al. 2014) and imageJ (Schneider et al. 2012) software packages. Changes in fluorescence were quantified by calculating F/F ₀, where F ₀ is the baseline fluorescence and F is the fluorescence during neuronal activity (both corrected for autofluorescence). Autofluorescence was estimated in a region of the same size of the region of interest and subtracted using a modified version of the ‘BG subtraction from ROI’ plugin by Dr M. Crammer. Slow experimental drift of the signal due to bleaching was subtracted after appropriate fitting of baseline fluorescence (Clampfit, Molecular Devices, San Jose, CA, USA).

Electrophysiological recordings and analysis

Slices were transferred to a direct microscope (Slicescope, Scientifica, Bellbrook Industrial Estate, Uckfield, UK) with oblique illumination Olympus optics (Olympus, Tokyo, Japan) and a Zyla sCMOS 4.2 camera. Cells were visualized using an Olympus 60× IR water immersion objective. Slices were superfused with preheated ACSF of the following composition (in mM): 130 NaCl, 24 NaHCO₃, 3.5 KCl, 1.25 NaH₂PO₄, 2 CaCl₂, 1 MgSO₄, 10 glucose, saturated with 95% O₂, 5% CO₂ at pH = 7.4 and maintained at a constant temperature (31–33°C) by a temperature controller (TC‐324B, Warner Instruments). CRs were recognized in the living slice by their location, typical tadpole shape, and tdTomato or EGFP fluorescence in tissue prepared from Wnt3a‐tdTomato and CXCR4‐EGFP animals, respectively. In experiments requiring ACSF with elevated external potassium concentrations, KCl was increased from 3.5 to 133.5 mM and NaCl was omitted.

Recording pipettes were pulled from borosilicate glass capillaries (Prism FLG15, Dagan Corporation, Minneapolis, MN, USA) and had a resistance of 3–5 MΩ when filled with the appropriate internal solution, as reported below. Recordings were performed using a Multiclamp 700 amplifier (Molecular Devices). The signals were filtered at 3 KHz and digitized at 20 kHz using a Digidata 1322A and the Clampex 9 program suite (Molecular Devices). Voltage‐clamp recordings were performed at the holding potential (V _h) indicated in the text.

In general, measurements and quantification of electrophysiological recordings were analysed using the Clampfit 9.0 (Molecular Devices), Origin Pro 7 (OriginLab, Northampton, MA, USA), Prism 6 (GraphPad Software, La Jolla, CA, USA), and Microsoft Excel suites of programs. In more detail, spontaneous synaptic events in interneurons were detected using the template‐based search algorithm of Clampfit (Molecular Devices). The time course of their frequency before and after exposure to capsaicin was calculated in bins of 5 s duration, and then normalized to the 30 s baseline preceding the application of capsaicin.

The actual (non‐normalized) frequencies given in the text and plotted in the figures reflect the value of the 30 s bin that mostly deviated from the baseline frequency measured during the 30 s preceding capsaicin application.

The effect of capsaicin on synaptic transmission was compared in different cell types by using the 30 s bin with the highest deviation from the baseline frequency after baseline normalization.

Pipette solutions

The composition of the solution used for voltage‐clamp recordings from CRs and interneurons was (in mM): 125 caesium methansulfonate, 0.3 GTP‐Na, 4 ATP‐Mg₂, 16 KHCO₃, 10 QX314‐Cl and 0.3–0.5% biocytin, equilibrated with 95% O₂, 5% CO₂ to a pH = 7.3. The solution used for current‐clamp recordings from CRs was (in mM): 105 potassium methylsulfate, 20 KCl, 10 NaCl, 0.3 GTP‐Na, 4 ATP‐Mg₂, 16 KHCO₃ and 0.3–0.5% biocytin, equilibrated with 95% O₂, 5% CO₂ to a pH = 7.3.

Statistics

Values in the text and data points ± bars shown in the time course plots of the effect of capsaicin application onto interneurons are means ± SE. Population box plots show the median as middle dash, lower and upper quartiles as upper and lower box borders, and minimum and maximum values as whiskers.

Statistical comparisons used non‐parametric tests and were performed with GraphPad Prism 6 software. Significance was accepted at the level of P < 0.05. Probability values in the text are rounded to the second decimal place.

When comparing more than two groups we relied on the Kruskal‐Wallis test with Dunn's multiple comparison post hoc test. When comparing two groups we used the Mann‐Whitney test and the Wilcoxon matched‐pairs signed rank test for unpaired and paired groups, respectively.

The statistical significance of the effect of capsazepine and Ruthenium Red in preventing capsaicin‐induced calcium responses was evaluated by calculating the probability of observing no responses in the six imaged cells (assumed as independent events) for each condition, with the probability of observing a response in a single cell (P _r) estimated as 0.5 (see Fig. 6), which yields a value of (1 − P _r)⁶ = 0.5⁶ = 0.02.

Figure 6. Responses to capsaicin are not present in every CR.

A, example of a microscopic field in control (upper panel), presence of capsaicin (middle panel) and elevated [K⁺]_o plus TTX (lower panel). Notice the presence of CRs not responsive to capsaicin. B, identical to A but with only capsaicin‐responsive CRs. Colour scale bar in arbitrary units for A and B. C, left panel, F/F ₀ of a CR not responsive to capsaicin, identified by the subsequent application of high potassium plus TTX. Right, similar to left panel, but for a capsaicin responsive CR.

Drugs

D‐AP5, NBQX, QX314‐Cl, and TTX were purchased from Alomone Labs (Jerusalem, Israel). Capsaicin, capsazepine, picrotoxinin, and Ruthenium Red were obtained from Sigma Aldrich (St Louis, MO, USA).

Results

In order to assess the presence of functional TRPV1 in CRs, we took advantage of the genetically encoded indicator GCaMP6s and calcium imaging (Fig. 1). First, we verified that in slices prepared from Wnt3a‐GCaMP6s mice (see Methods for details) GFP‐immunoreactive cells were mostly found in the molecular layers of the hippocampus, and were especially abundant in the CA3 subfield (Fig. 1 A), consistent with the quantitative description of hippocampal CRs provided by Anstötz et al. (2016). More importantly, when observed at high magnification, labelled cells of the molecular layers had the typical ‘tadpole‐like’ morphological features of CRs, with a single main dendrite emerging from the soma opposite to the axon initial segment (Fig. 1 B; compare to: von Haebler et al. 1993; Marchionni et al. 2010; Quattrocolo & Maccaferri, 2013, 2014; Anstötz et al. 2016). We also performed direct electrophysiological recordings from n = 10 GCaMP6s‐expressing cells in the molecular layers. Their responses to depolarizing current steps consisted of action potentials of declining amplitude followed by membrane oscillations (Fig. 1 C), and their estimated membrane resistance yielded high values (2.5 ± 0.3 GΩ), similar to those previously reported for CRs (Marchionni et al. 2010; Quattrocolo & Maccaferri, 2013, 2014; Anstötz et al. 2016). Lastly, in n = 6 cases, biocytin filling of the recorded cells allowed us to recover their morphology and unequivocally confirm their anatomical identity (Fig. 1 D).

Figure 1. Expression of the genetically encoded calcium indicator GCaMP6s in hippocampal CRs.

A, confocal image showing GFP immunostaining in a transverse hippocampal section of a P12 Wnt3a‐GCaMP6s mouse. Notice the typical distribution of CRs in the hippocampal molecular layers. Dotted line, hippocampal fissure. The continuous line delimitates the hippocampal regions: Cornu Ammonis (CA) 1, CA 3, dentate gyrus (DG), and subiculum (Sub). Box outlines the field of higher magnification shown in B. B, higher magnification image showing the typical tadpole‐like morphology of CRs. C, micrograph of a biocytin‐filled GCaMP6s‐expressing cell showing the characteristic morphological properties of CRs. D, GCaMP6s‐expressing neurons show the typical firing pattern, sag and high membrane resistance described in CRs. Traces are responses to −25/+60 pA current steps.

Next, we verified that the level of expression of the genetically encoded indicator was sufficient to reveal functional calcium responses by exposing slices to two experimental conditions that are predicted to increase intracellular calcium concentrations (Fig. 2). First, we tested the effects of the potassium channel blocker 4‐aminopyridine (4AP, which triggers spontaneous epileptiform bursting in CRs; see Marchionni et al. 2010), followed by the exposure to elevated external potassium concentrations (high K⁺ in the presence of TTX, in order to depolarize the membrane potential). In 79% of n = 38 cells, we could observe both 4AP‐induced calcium waveforms and responses to high K⁺, whereas the remaining cells only displayed responses to high K⁺. The most parsimonious explanation for the fraction of CRs not responding to 4AP is that some cells have severed input connections and/or dendritic trunks following the slicing procedure, which would exclude them from synaptically driven activity generated by 4AP.

Figure 2. Calcium responses of CRs in slices from Wnt3a‐GCaMP6s mice exposed to 4‐aminopyridine (4AP) and high external potassium ([K⁺]_o = 133.5 mM) plus TTX.

A, pseudo‐coloured averaged images collected at different times during application of 4AP (red bars marked ‘a’ and ‘b’ in B) and in the presence of elevated [K⁺]_o plus TTX (cyan bar marked ‘c’ in B). Colour scale bar at the right (arbitrary units). B, notice the periodic increases in normalized fluorescence (F/F ₀) during addition of 4‐aminopyridine and the large response following the direct depolarization induced by exposure to high [K⁺]_o in the presence of TTX.

In summary, these initial experiments indicate that the Wnt3a‐GCaMP6s mouse is an appropriate tool for the study of calcium responses in hippocampal CRs. Therefore, we decided to take advantage of this functional approach to study TRPV1‐dependent responses.

When exposed to the TRPV1 agonist capsaicin (Caterina et al. 1997) at 10 μM, 1 μM, and 100 nM, some CRs revealed clear, but variable, increases in intracellular calcium (Fig. 3). Pseudo‐coloured images for the effect of capsaicin at different concentrations are shown in Fig. 3 A, whereas their individual quantification and summary plots are illustrated by Fig. 3 B and C, respectively. Peak F/F ₀ was 13.4 ± 2.3 for 10 μM (n = 9 cells), 17.0 ± 2.2 for 1 μM (n = 8 cells) and 14.4 ± 1.7 for 100 nM (n = 5 cells). The half‐width of the response was 123 ± 16 s for 10 μM (n = 7 cells), 118 ± 14 s for 1 μM (n = 8 cells) and 116 ± 33 s for 100 nM (n = 4 cells). The latency to the main observed calcium response was long and dose dependent, with values of 132 ± 9 s for 10 μM (n = 9 cells), 145 ± 14 s for 1 μM (n = 8 cells), and 275 ± 60 s for 100 nM (n = 5 cells). The latencies of responses to micromolar capsaicin concentrations (10 μM and 1 μM) were shorter than that measured for 100 nM (P < 0.05 in both cases). These results are consistent with the slow diffusion of capsaicin from the extracellular space to its TRPV1 intracellular binding site (Cao et al. 2013).

Figure 3. The TRPV1 activator, capsaicin, triggers calcium responses in hippocampal CRs.

A, pseudo‐coloured average images collected in the absence (top panels) and in the presence of the drug at different concentrations (lower panels). Colour scale bar at the right (arbitrary units). Notice the typical tadpole‐like appearance of the imaged cells. B, upper panels, calcium responses to different capsaicin concentrations (left to right: 10 μM, 1 μM and 100 nM) from different cells shown in different colours and superimposed. Notice the large variability. Lower panels, same records as above, but normalized for amplitude and corrected for the different latencies show a relatively similar kinetics. C, quantification of peak amplitudes, half‐width and latencies of the F/F ₀ responses. Notice the significant concentration dependency of the latencies, and the large variability in amplitudes and half‐widths.

Although these results show that application of capsaicin triggers calcium responses in CRs, the site of action of capsaicin remains potentially undetermined. In fact, while Cavanaugh et al. (2011) have suggested that CRs are the only hippocampal neuronal type expressing TRPV1, other studies indicate a more widespread expression, including several other neuronal types (i.e. pyramidal cells, GABAergic interneurons, and granule cells; see Toth et al. 2005; Cristino et al. 2006; Gibson et al. 2008; Chávez et al. 2010, 2014; Puente et al. 2015; Canduela et al. 2015; Lee et al. 2015; Eguchi et al. 2016; Hurtado‐Zavala et al. 2017; see also review by Kauer & Gibson, 2009). Thus, application of capsaicin might trigger some form of network activity, which would result, indirectly, in responses in CRs. In addition, as TRPV1 activation would be predicted to generate inward currents at resting membrane potentials, we could not exclude that the calcium signals observed in CRs were mostly due to membrane depolarization and opening of voltage‐dependent calcium channels. Therefore, in order to distinguish among these possibilities, we took advantage of pharmacological manipulations that inhibit postsynaptic activity, action potential generation, and block voltage‐dependent calcium channels (Fig. 4). Figure 4 A shows pseudo‐coloured images illustrating capsaicin‐induced responses during the three aforementioned manipulations. Quantification of responses from individual CRs is shown in Fig. 4 B, and summary plots in Fig. 4 C.

Figure 4. Responses to capsaicin are not dependent on the integrity of synaptic transmission and/or action potential generation.

A, left panel, control image in the presence of antagonists of synaptic ionotropic receptors (syn blockers; NBQX, 20 μM; D‐AP5, 50 μM; picrotoxinin, 50 μM) and after the addition of capsaicin (1 μM). Middle panel, similar to left panel with the additional constant presence of TTX. Right panel, control image in the presence of TTX and cadmium (150 μM) and after the introduction of capsaicin (1 μM). Notice that none of the experimental conditions was successful in preventing responses to capsaicin. Colour scale bar in arbitrary units. B, F/F ₀ traces obtained from several CRs shown in different colours and superimposed in the same experimental conditions of the panels in A. C, summary plots from several experiments quantifying the peak of the F/F ₀ response (left), its half‐width (middle) and latency (right).

In the first set of experiments we exposed slices to capsaicin (1 μM) in the constant presence of a cocktail of synaptic receptor blockers (picrotoxinin, 100 μM; NBQX, 20 μM; D‐AP5, 50 μM). Despite pharmacological blockade of the network, calcium responses could still be detected, thus suggesting that TRPV1 expressed by CRs are indeed a direct target of capsaicin, and that no synaptically driven activity is required. Next, we repeated the experiment in the additional presence of TTX (500 nM) in order to exclude action potential‐induced calcium entry, which may be triggered in CRs by TRPV1‐dependent depolarization. Under these experimental conditions, we were still able to observe responses, indicating that firing in CRs is not a necessary requirement either. Lastly, we applied capsaicin in the simultaneous presence of TTX (500 nM) and cadmium (150 μM) in order to test the contribution of voltage‐dependent calcium channels that could be activated in CRs by sub‐threshold depolarizations (Kirmse et al. 2005). Similar to what was observed in the previous experiments, responses were not prevented, but still persisted. Taken together, these results suggest that the critical element leading to the increase in intracellular calcium concentrations is the calcium flux mediated by TRPV1 expressed by CRs themselves. This conclusion is also consistent with the different profile of the responses observed during capsaicin application compared to 4AP‐triggered network activity. In fact, peak F/F ₀ (12.6 ± 2.8 in synaptic blockers, n = 8 cells; 11.4 ± 2.3 in synaptic blockers and TTX, n = 9 cells; 15.1 ± 3.6 in TTX and cadmium, n = 10 cells), half‐width durations (136 ± 18 s in synaptic blockers, n = 8 cells; 163 ± 26 s in synaptic blockers and TTX, n = 8 cells; 94 ± 26 s in TTX and cadmium, n = 10 cells), and latency to the response (119 ± 14 s in synaptic blockers, n = 8 cells; 116 ± 11 s in synaptic blockers and TTX, n = 9 cells; 108 ± 10 s in TTX and cadmium, n = 10 cells) appeared very similar under all these experimental conditions.

The efficacy of the aforementioned manipulations was directly confirmed in a set of electrophysiological measurements (Fig. 5). First, application of the cocktail of synaptic blockers completely suppressed spontaneous synaptic currents observed in CRs (Fig. 5 A, from a frequency of 0.3 ± 0.1 Hz in the absence of the drugs to 0.0 ± 0.0 in their presence, n = 7 cells, P = 0.02). Second, TTX blocked current‐evoked action potentials (Fig. 5 B, from 9 ± 5 in control conditions to 0 ± 0 in the presence of the toxin, n = 8 cells, P = 0.01). Third and last, application of cadmium in the continuous presence of TTX eliminated a sub‐threshold, current‐evoked, active voltage‐response (Fig. 5 C). Because of the presence of membrane oscillations that prevented a precise measurement of the membrane potential, this difference was quantified by subtracting the area of the voltage response during the current step in control from that recorded in the presence of cadmium. The measured values were reduced from 22 ± 1 mV s to 18 ± 2 mV s in the absence vs. the presence of cadmium (n = 6 cells, P = 0.03), yielding a cadmium‐sensitive area of 4 ± 1 mV s.

Figure 5. Electrophysiological validation of the effectiveness of the drugs used in the calcium imaging experiments of Fig. 4 .

A, complete blockade of spontaneous synaptic events (control, upper traces) following the exposure to the synaptic blocker cocktail (NBQX, 20 μM; D‐AP5, 50 μM; picrotoxinin, 50 μM; lower traces). Five sweeps superimposed in both conditions: a single trace is highlighted in red. B, suppression of current‐induced firing in CR (control, upper trace) by application of TTX (0.5 μM, lower trace). The inset (red) shows typical membrane oscillations remaining in the presence of TTX. Positive current step, 50 pA. C, effect of cadmium (150 μM) on current‐induced depolarizations in the presence of TTX. Upper panel, notice the smaller responses recorded after cadmium addition, consistent with the blockade of a voltage‐dependent depolarizing calcium current (five sweeps superimposed in control as black traces and in the presence of cadmium as red traces). Lower panel, averaged traces (5 sweeps) in the two different conditions (black, control; red, cadmium) are superimposed to show the quantification of cadmium effect as the difference in the area identified by the voltage responses in the two conditions (grey area). Current step, 25 pA.

Next, in order to gain insights into the percentage of CRs producing capsaicin‐triggered responses, we designed a protocol where an initial application of the agonist was followed by exposure to high K⁺ in the continuous presence of TTX (Fig. 6). From a sample of n = 34 cells responding to high K⁺, 50% of the tested neurons were not affected by capsaicin (Fig. 6 A), whereas the remaining 50% produced calcium elevations both in the presence of capsaicin and during exposure to high K⁺. This result suggests either that half of the CR population expresses TRPV1 or that the TRPV1 conductance may be desensitized and/or degraded in some cells (Sanz‐Salvador et al. 2012).

Although capsaicin is considered a classical agonist of TRPV1, it has also been proposed that it may impact other, still undetermined, signalling pathways (Benninger et al. 2008). Therefore, as shown in Fig. 7, we tested the sensitivity of capsaicin‐induced responses to pharmacological blockers with distinct modes of action and specificity (Vriens et al. 2009), i.e. the TRPV1‐selective competitive antagonist capsazepine (10 μM, Fig. 7 A), and the broad non‐competitive antagonist Ruthenium Red (20 μM, Fig. 7 B). In the presence of either of the drugs, application of capsaicin was ineffective and could not generate any type of response (capsazepine, n = 6 cells, P = 0.02, and Ruthenium Red, n = 6 cells, P = 0.02).

Figure 7. TRPV1 antagonists prevent capsaicin‐induced calcium responses in CRs.

A, left panel, in the constant presence of capsazepine (10 μM), application of capsaicin (1 μM) does not generate responses. Different CRs are identified by traces of different colours. Right panel, verification that the same cells could generate calcium responses when exposed to 4‐aminopyridine (100 μM). B, identical to A, but the broad TRPV‐broad spectrum antagonist Ruthenium Red (20 μM) was used to prevent responses to capsaicin (10 μM).

Thus, the data presented so far argue for the presence of functional TRPV1 in CRs. However, they do not clarify whether calcium responses are triggered by TRPV1 expressed on the plasma membrane vs. the membrane of intracellular calcium stores (Kárai et al. 2004; Lee et al. 2015).

We reasoned that if CRs expressed TRPV1 on the membrane of their intracellular calcium stores, we should be able to observe responses to capsaicin in the absence of calcium from the external solution. In contrast, if TRPV1 were mainly expressed on the plasma membrane, capsaicin‐triggered responses should critically depend on the presence of calcium in the external medium. When capsaicin was applied in a modified ACSF containing nominal zero calcium and 150 μM EGTA, very low‐amplitude responses were observed (Fig. 8 A, left). However, the same cells revealed clear and much larger amplitude responses when a regular calcium‐containing external solution was introduced (Fig. 8 A, right). Peak F/F ₀ values during capsaicin application in zero calcium/EGTA ACSF were 1.5 ± 0.2, which compared to 8.0 ± 1.5 in standard ACSF, n = 6 cells, P = 0.03. As a control experiment, we measured the impact of the same switch from nominal zero calcium/150 μM EGTA to regular calcium (2 mM) ACSF in the absence of capsaicin (Fig. 8 B, left). Only a modest, but significant increase in baseline levels was observed and peak F/F ₀ values increased from 1.1 ± 0.0 in zero calcium/EGTA ACSF to 1.3 ± 0.1 in regular ACSF (n = 17, P = 0.002). When the same cells were later exposed to high K⁺ solution, calcium responses of an order of magnitude larger could be clearly observed (peak F/F ₀: 11.4 ± 0.9, n = 17), thus indicating that the imaged neurons were capable of producing large responses (Fig. 8 B, right). Taken together, these results suggest that the TRPV1 channels that mostly contribute to the response are primarily located in the plasma membrane.

Figure 8. External calcium influx mediates responses to capsaicin.

A, top panels, pseudo‐coloured averaged images of a CR in zero calcium/EGTA external solution (0 Ca⁺² EGTA) before (left panel) during the addition of capsaicin 10 μM, middle panel) and after the reintroduction of calcium (2 mM) and removal of EGTA (right panel). Colour scale bar in arbitrary units. Bottom panel, quantification of F/F ₀ showing the dependence of capsaicin‐induced responses on external calcium. Results from several cells are shown by traces of different colours and superimposed. Notice that near‐absence of responses to capsaicin in zero calcium/EGTA external solution (left traces) and the reappearance of a response when EGTA is removed and external calcium is reintroduced in the same cells (right traces). B, left panel, reintroduction of calcium in the external solution generates slightly increased baseline levels, which are much smaller than responses observed in the presence of capsaicin. Right panel, same cells exposed to elevated [K⁺]_o. The different timing of the calcium response is due to slight differences in the application of the high K⁺ external solution. The black bar indicates the earliest time of application.

In order to further verify this interpretation, we recorded TRPV1‐mediated currents from visually identified CRs (Wnt3a‐tdTomato mice) kept under whole‐cell configuration. In n = 6 out of 9 cells, inward currents were observed following capsaicin application (10 μM) at a holding potential of −60 mV (Fig. 9 A, left). The average calculated charge transfer was 4.0 ± 2.1 nC. In contrast, when capsaicin was used in slices constantly exposed to capsazepine (100 μM), currents were never detected, and the estimated charge transfer was (0.0 ± 0.1 nC n = 6 cells, Fig. 9 A, right). In a different set of experiments, we compared the average charge transfer following capsaicin application in visually identified CRs of CXCR4‐EGFP‐TRPV1^+/+ vs. CXCR4‐EGFP‐TRPV1^−/− mice (Fig. 9 B). When recordings were made from slices obtained from TRPV1^+/+ animals, capsaicin triggered an inward current in 6 out of 14 recordings (Fig. 9 B, left), whereas no current was ever identified in CRs of TRPV1^−/− mice (n = 22 cells, Fig. 9 B, right). The charge transfer calculated from all the experiments performed in the TRPV1^+/+ group (including apparently non‐responding CRs) was significantly different from that estimated when the experiment was performed in slices prepared from the TRPV1⁻/⁻ genotype (1.0 ± 0.4 nC, n = 14 cells vs. 0.2 ± 0.1 nC, n = 22 cells, P = 0.04). When non‐responding cells of the TRPV1^+/+ group were excluded from the analysis, the charge transfer increased to 2.0 ± 0.7 nC, (n = 6 cells), which was not significantly different from the results described previously in Wnt3a‐tdTomato mice (P = 0.81). Fig. 9 C shows confocal images of typical tdTomato‐ and EGFP‐expressing CRs obtained from the three mouse lines used for these experiments.

Figure 9. TRPV1‐mediated currents can be recorded in whole‐cell conditions from CRs.

A, left top panel, voltage‐clamp recording (V _h = −60 mV) from a CR during capsaicin application (10 μM). Notice the development of a slow inward current. Bottom panel, summary plot of the current time course obtained by averaging capsaicin responsive cells (means ± SE). Right top and bottom panels, same experiment, but in the constant presence of capsazepine (100 μM). In the presence of the antagonist, capsaicin did not induce any inward current. Wnt3a‐td‐tomato reporter mice were used to facilitate the identification of CRs. B, left top panel, TRPV1‐mediated current in a CR recorded in a slice obtained from a CXCR4‐EGFP‐TRPV1⁺/⁺ mouse. Notice the increase in noise in the presence of capsaicin. Bottom panel, summary graph of capsaicin‐induced currents for all CRs tested (cyan circles) and for capsaicin‐responsive cells only (blue circles). Notice the presence of an inward current in both type of analysis. Right panels, same experiments as in the left panel, but performed in slices from CXCR4‐EGFP‐TRPV1⁻/⁻ animals. Notice the complete absence of capsaicin‐induced currents both in the individual example (top trace) and in the population analysis (bottom graph). C, confocal images showing tdTomato (top panel) and EGFP (middle and bottom panel) labelling of CRs in the hippocampal molecular layers of the mice strains used for the experiments. Notice the typical unipolar shape of CRs shown in more detail in the bottom right insets.

In summary, our results provide compelling evidence supporting the expression of functional TRPV1 on the plasma membrane of hippocampal CRs, but do not clarify their potential roles and impact on the network.

Several groups have provided evidence for network‐related effects following the activation of TRPV1 in the hippocampus. For example, TRPV1 expressed by postsynaptic dentate granule cells has been shown to control synaptic plasticity of medial perforant path excitatory inputs (Chávez et al. 2010), and to reduce perisomatic GABAergic inputs (Chávez et al. 2014). In the hippocampus proper, Gibson et al. (2008) have reported that the stimulation of TRPV1 located on the presynaptic terminals of the Schaffer collaterals decreases evoked excitatory synaptic transmission on stratum radiatum interneurons, and induces long‐term depression. More recent work has also proposed that TRPV1 expressed by hippocampal O‐LM interneurons of stratum oriens promotes the formation of excitatory synapses on this cell type (Hurtado‐Zavala et al. 2017). Notably, the impact of TRPV1 activation on the microcircuits of the molecular layers innervated by CRs (either in the dentate gyrus or on the Cornu Ammonis) has not been examined yet. Therefore, given our evidence that CRs express functional TRPV1, we decided to test the prediction that the pharmacological stimulation of slices with capsaicin would produce changes in synaptic activity of CR targets, such as GABAergic interneurons of the molecular layers (Figs 10, 11, 12 and 13).

Figure 10. TRPV1‐dependent enhancement of the frequency of spontaneous excitatory events in hippocampal interneurons of the molecular layers.

A, left panel, application of capsaicin (10 μM) increases the occurrence of glutamatergic events recorded at −60 mV in interneurons of the molecular layers in slices from CXCR4‐EGFP‐TRPV1⁺/⁺ mice. Summary plot of the time course of the effect of capsaicin on the normalized frequency of the events (norm event freq) and an individual recording are shown at the bottom and top, respectively. Right panel, graph of the actual frequency values in control and in the presence of capsaicin for all the individual experiments and box plots for each condition. Insets show five superimposed sweeps in control (lighter traces) and in the presence of capsaicin (darker traces). Notice the increased occurrence of spontaneous events. B, same experiments as in A performed in CXCR4‐EGFP‐TRPV1⁻/⁻ animals. Notice the lack of effect of capsaicin, indicating that the results observed in A are mediated by TRPV1.

Figure 11. The effect of TRPV1 activation on excitatory synaptic transmission does not require action potential generation in the network.

A, left panel, capsaicin (10 μM) superfusion enhances the frequency of miniature glutamatergic events recorded in interneurons of the molecular layers (V _h = −60 mV) in the constant presence of TTX (0.5 μM throughout). Slices from CXCR4‐EGFP‐TRPV1⁺/⁺ mice. Summary plot from several experiments and an individual recording are shown at the bottom and top, respectively. Right panel, graph of the actual frequency values in control and in the presence of capsaicin for all the single recordings in control and after the addition of capsaicin, and box plots for the two conditions. Insets show five superimposed sweeps in control (lighter traces) and in the presence of capsaicin (darker traces). B, same experiments as in A performed in CXCR4‐EGFP‐TRPV1⁻/⁻ animals. Notice the lack of effect of capsaicin, indicating that the results observed in A are mediated by TRPV1.

Figure 12. TRPV1 activation has a much stronger impact on spontaneous synaptic activity of interneurons of the molecular layers compared to horizontal cells of stratum oriens.

Upper traces show individual examples of recordings from interneurons located in stratum lacunosum‐moleculare (blue trace) and stratum oriens (red trace). Notice the larger increase of postsynaptic currents in the stratum lacunosum moleculare (l‐m) interneuron. V _h = −60 mV for both experiments. The time course of the effect of capsaicin application on postsynaptic currents recorded is shown in the left lower graph, whereas the summary box plots for the different cell types are shown in the right lower panel (interneurons of the molecular layers. blue symbols; interneurons of stratum oriens, red symbols).

Figure 13. Anatomical reconstruction and post hoc superimposition in an idealized slice of the interneurons that were recovered after the experiments of Fig. 12 .

Data are from n = 7 interneurons of the molecular layers and n = 7 stratum oriens interneurons; white circles identify their soma. Notice the layer‐specific restricted distribution of the dendritic (red) and axonal arborizations (blue) to the molecular layers (stratum lacunosum‐moleculare, SLM; molecular layer of the dentate gyrus, ML), in one group (interneurons of the molecular layers), and the confined localization of the dendrites of the group of ‘horizontal interneurons’ to stratum oriens (orange), with axonal projections (green) to various layers (stratum oriens. SO; stratum pyramidale, SP; stratum radiatum, SR). GCL, granule cell layer.

When we exposed slices prepared from CXCR4‐EGFP‐TRPV1^+/+ animals to capsaicin, we noticed a large increase in the frequency of spontaneous excitatory postsynaptic currents (sEPSC) in voltage‐clamped interneurons of the molecular layers (1.1 ± 0.2 Hz in control vs. 8.9 ± 1.7 Hz in the presence of capsaicin, n = 11 cells, P = 0.001; Fig. 10 A). The effect of capsaicin appeared to be specific for TRPV1, as it was absent in tissue obtained from CXCR4‐EGFP‐TRPV1^−/− animals (1.4 ± 0.4 Hz in control vs. 1.4 ± 0.3 Hz in the presence of the drug, n = 11 cells, P = 0.67; Fig. 10 B). In additional experiments, we verified that application of NBQX completely eliminated spontaneous events recorded in the presence of capsaicin, thus confirming that they were glutamatergic excitatory postsynaptic currents (n = 8 cells; data not shown).

Next, we wondered whether capsaicin‐induced increase in spontaneous glutamatergic transmission was dependent on action potential‐driven release of neurotransmitter or could be still observed in the presence of TTX (Fig. 11). Capsaicin greatly increased the frequency of miniature excitatory postsynaptic currents (mEPSCs) in slices prepared from CXCR4‐EGFP‐TRPV1^+/+ (from 1.0 ± 0.2 Hz in control to 11.5 ± 2.1, n = 10 cells, P = 0.002; Fig. 11 A), but not from CXCR4‐EGFP‐TRPV1^−/− animals (0.9 ± 0.3 Hz vs. 1.0 ± 0.2, n = 8 cells, P = 0.89; Fig. 11 B). Taken together, these data indicate that an increase in action potential frequency does not mediate the observed network effect of capsaicin in the molecular layers of the hippocampus, but suggest a direct effect on the release machinery of the terminals.

As a capsaicin has been recently proposed to enhance excitatory synaptic transmission onto stratum oriens interneurons via TRPV1 located on their postsynaptic membranes (Hurtado‐Zavala et al. 2017), we decided to compare the impact of TRPV1 activation onto interneurons located in the molecular layers vs. ‘horizontal’ interneurons stratum oriens (Maccaferri et al. 2000; Maccaferri, 2005).

Thus, we decided to compare the degree of potentiation of sEPSC frequency in interneurons with dendritic arborization mostly confined to the molecular layers vs. stratum oriens. The effect of capsaicin on sEPSC frequency was much larger in cells located in the molecular layers, although a small acute effect was observed also in stratum oriens neurons. Fig. 12 shows the quantification of the results for the two different subsets of interneurons. The ratio of sEPSC frequency in capsaicin over control was 10.0 ± 1.6 in seven cells located in the molecular layers vs. 2.9 ± 1.3, in eleven cells of stratum oriens (P = 0.002). The anatomical reconstructions of the recorded cells (7/7 cells in the molecular layers and 7/11 cells of stratum oriens) show very clearly the segregation of the dendritic arbors to the layers hosting the cell body (Fig. 13). Notice, also, the dense and local axonal arborization of the multipolar, stellate‐like interneurons of the molecular layers (typical of neurogliaform cells; see Overstreet‐Wadiche & McBain, 2015) compared to variable axonal projections of ‘horizontal’ neurons of stratum oriens (Maccaferri, 2005). In conclusion, acute pharmacological TRPV1 activation appears to predominantly control synaptic events in hippocampal microcircuits of the molecular layers. Furthermore, as these last experiments were performed in wild‐type C57BL/6J animals, the observed effects of capsaicin on synaptic transmission do not depend on unpredictable mechanisms potentially occurring in transgenic animals.

Discussion

This work establishes that CRs express functional TRPV1 on their plasma membranes, and that the activation of TRPV1 in hippocampal slices has a strong impact on synaptic transmission in the molecular layers.

Functional TRPV1 is expressed in CRs

Our data conclusively demonstrate the validity of the original report by Cavanaugh et al. (2011) that hippocampal CRs express TRPV1. Furthermore, we expand that finding by showing that TRPV1s in CRs are fully functional and are located predominantly, if not exclusively, on CR plasma membrane. We also suggest that TRPV1 on CR terminals powerfully controls glutamate release onto postsynaptic GABAergic targets. The physiological relevance of TRPV1 activation at synaptic boutons can be easily related to its calcium permeability, as presynaptic calcium concentration controls vesicle fusion and neurotransmitter release (Dodge & Rahamimoff, 1967).

However, it is important to acknowledge that, in contrast to the direct approach (i.e. calcium imaging and electrophysiological recordings from CRs) used to conclude that CRs express functional TRPV1, our proposal of a TRPV1 role in synaptic release is based on more indirect reasoning. In fact, we think that a direct experimental test of this hypothesis is prevented by technical constraints.

A direct comparison of CR‐evoked postsynaptic currents in interneurons in the absence vs. presence of capsaicin would necessitate the ability to collect and average a very large number of events. This would be required by the massive increase of the noise generated by the background spontaneous events in the presence of capsaicin. Under our reported experimental conditions, optogenetic activation of CRs is unfortunately limited by the long inter‐flash interval required to ensure their reliable firing (between 30 s and 1 min; see Quattrocolo & Maccaferri, 2014 and Anstötz et al. 2018 for further details). Hence, this experimental approach would imply unrealistically long experiments. Furthermore, prolonged exposure to capsaicin would make interpretations difficult because of agonist‐ and calcium‐dependent desensitization of TRPV1 channels (Sikand & Premkumar, 2007; Sanz‐Salvador et al. 2012).

An alternative strategy based on direct paired recordings is also unrealistic due to the very low probability of finding connections in slices (Anstötz et al. 2016). Additionally, paired recordings would generate unitary postsynaptic currents of smaller amplitude than optogenetically stimulated compound events. Therefore, in order to compensate for a capsaicin‐induced increase in background noise, an even larger numbers of events and even longer experiments would be required. Lastly, although a genetic approach based on the selective removal of TRPV1 from CRs would be optimal, TRPV1‐floxed animals are not available at present.

Alternative interpretations are indeed possible, as TRPV1‐expressing neurons have been reported in the entorhinal cortex by Cavanaugh et al. (2011). If these cells were glutamatergic neurons originating in the perforant path and/or temporoammonic pathway, then the possibility of TRPV1 expression on their terminals mediating the observed effect on synaptic transmission should not be excluded. However, direct recordings from layer II/III entorhinal cortex neurons, which originate in the perforant path and temporoammonic pathway (Steward & Scoville, 1976; van Groen et al. 2003; Kitamura et al. 2014), have only revealed TRPV1‐dependent depression of evoked inhibitory transmission, with no additional effects on excitatory postsynaptic currents (Banke, 2016). These changes were interpreted as indicative of presynaptic TRPV1 on local GABAergic terminals, because they were associated with increased paired pulse modulation and decreased CV² values (Banke, 2016). Therefore, in the absence of any evidence supporting functional TRPV1 expression in glutamatergic neurons of the entorhinal cortex, our current hypothesis may provide the most parsimonious explanation. Future work with the selective removal of TRPV1 on CRs will provide conclusive supporting (or opposing) evidence.

Layer‐specific impact of TRPV1 activation in the hippocampus

Irrespective of the caveats and technical limitations mentioned above, our data reveal a striking uniqueness of TRPV1‐dependent effects on excitatory transmission onto GABAergic interneurons of the molecular layers compared to what has been reported in other hippocampal cell types located in stratum granulosum, radiatum and oriens. Capsaicin generates TRPV1‐dependent depression of excitatory inputs both in dentate granule cells and stratum radiatum interneurons, albeit with important differences (dependent on postsynaptic TRPV1 in granule cells (Chávez et al. 2010) and on presynaptic TRPV1 in stratum radiatum interneurons (Gibson et al. 2008)). Here, we report TRPV1‐dependent modulation of excitatory synaptic transmission in interneurons of the molecular layers in the opposite direction, i.e. a massive increase of spontaneous release.

Although the acute effects of TRPV1 stimulation on synaptic transmission were not studied, Hurtado‐Zavala et al. (2017) have recently reported that TRPV1 activation promotes the development of excitatory innervation of O‐LM interneurons. Similar to the interneurons recorded in this work, O‐LM interneurons target the molecular layers of the hippocampus proper, i.e. stratum lacunosum‐moleculare, where the distal dendrites of pyramidal cells are located (Maccaferri et al. 2000). This observation, together with our results and with the TRPV1‐dependent depression of excitatory drive on stratum radiatum interneurons (Gibson et al. 2008), suggests that the activation of TRPV1 in the hippocampal network may selectively enhance disynaptic GABAergic inhibition to the distal dendrites of pyramidal cells, while reducing it in the remaining somatic/proximal dendritic domains. Such a dynamic shift from predominantly perisomatic to distal dendritic inhibition has been observed during high‐frequency stimulation of inhibitory microcircuits in hippocampal slices in vitro (Pouille & Scanziani, 2004). This phenomenon is likely to underlie the gating of dendritic spikes observed in the hippocampus in vivo during cooperative network activity (Kamondi et al. 1998).

Although the activation state of TRPV1 in the hippocampus under specific network conditions remains undetermined, the availability of several putative physiological TRPV1 ligands is known to be promoted by specific types of excitatory afferent stimulations (theta burst for 12‐(S)‐HPETE, DeCostanzo et al. 2010; high‐frequency stimulation for anandamide, Chávez et al. 2010). It is also interesting to note that anandamide, which can act as a direct TRPV1 activator (Smart et al. 2000), is an agonist of the CB1 receptor (Felder et al. 1993), which reduces perisomatic GABAergic inhibition provided by cholecystokinin (CCK)‐expressing basket cells (Katona et al. 1999; Földy et al. 2006). Thus, increasing the activation state of TRPV1 in hippocampal circuits may provide a mechanism to alter the balance between distal vs. proximal dendritic/somatic GABAergic inhibition.

This type of postsynaptic domain‐specific inhibitory shift has been proposed by Paulsen & Moser (1998) to promote different computations occurring during what they termed ‘read‐in’ and ‘read‐out’ modes of the hippocampal network, which they associated with enhanced/depressed synaptic plasticity and backpropagating/axonal action potentials, respectively.

Additional considerations on potential cellular/network physiopathological roles of TRPV1 in CRs

Occasionally, we noticed abnormalities in some imaged CRs that could be interpreted as a sign of cell pathology during application of capsaicin (see for example, the dendritic beads of Fig. 3 A, middle panel). As TRPV1 forms calcium permeable channels, its activation has the potential to trigger cellular toxicity/cell death in neuronal and non‐neuronal cells (Kim et al. 2005; Thomas et al. 2007; Hu et al. 2008; Stock et al. 2012). Thus, we speculate that strong activation of TRPV1 could play a role in the cell death‐dependent regulation of the density of CRs in the hippocampal network (Anstötz et al. 2016, 2018). If TRPV1s were involved in this process, then our observation of a high degree of response variability to capsaicin, and even the presence of a large proportion of non‐responding cells, could be a marker of the specific dynamic position of a cell in the circuit. Strong functional TRPV1 responses could be the reflection of CRs ready to be eliminated, whereas non‐responding cells could be part of a persisting population. This hypothesis will remain speculative until the degree of TRPV1 activation can be assessed under physiological, rather than pharmacological (and hence artificial) conditions.

Lastly, it is intriguing to notice that patients suffering from temporal lobe epilepsy with Ammon's horn sclerosis and who had experienced complex febrile seizures at young ages have abnormally large numbers of CRs in the hippocampus (Blümcke et al. 1996, 1999). Although this finding may suggest a maldevelopmental disorder and the involvement of CRs in some form of temperature‐initiated epileptogenesis (Blümcke et al. 2002), recombinant TRPV1 channels are activated by heat with a threshold of about 45°C (Tominaga et al. 1998). This value may appear too elevated for febrile seizures. However, the temperature threshold of TRPV1 can be powerfully modulated by ligands and by the phosphorylation state of the channel (Benham et al. 2003). In conclusion, even if it cannot be excluded that under particularly unfavourable conditions TRPV1 activity in CRs may contribute to the pathophysiology of febrile seizures in a subpopulation of patients, additional work is required to conclusively prove or disprove this hypothesis.

Conclusions

Our data reveal the functional expression of TRPV1 in CRs and a striking TRPV1‐dependent layer‐specific modulation of synaptic transmission in the hippocampus. These findings are important to pursue a cell type‐specific understanding of the role of TRPV1 in the regulation of hippocampal microcircuits. We suggest that the activation of TRPV1 on CRs may have a profound impact on GABAergic input to the hippocampal molecular layers, which may be important for the physio‐pathological regulation of their local processing (Maccaferri, 2011).

Additional information

Competing interests

The authors have no competing interests.

Author contributions

G.M conceived and designed the experiments. M.A., S.K.L., and G.M. collected, analysed and interpreted the experimental data. G.M. drafted the article, and all authors contributed to writing the manuscript. All authors have approved the final version of the manuscript, and agree to be accountable for all aspects of the work.

Funding

This work was supported by the National Institute of Neurological Disease and Stroke (Grant NS064135 to G.M.).

Biography

Max Anstötz earned his MD at the RWTH‐Aachen University in Germany with a research focused on structural hippocampal neuronal diversity under the mentorship of Professor Joachim Lübke. He then moved to University Medical Center Hamburg‐Eppendorf under the supervision of Professor Gabriele Rune, where he studied cortical development in the Reeler mouse. Presently, he is a post‐doctoral fellow at the Feinberg Medical School, Northwestern University. He is interested in unrevealing the roles of hippocampal Cajal‐Retzius cells in physiopathological signal processing and integration.