Serotonin 5-HT₃ receptors in rat CA1 hippocampal interneurons: functional and molecular characterization

Abstract

The molecular makeup of the serotonin 5-HT₃ receptor (5-HT₃R) channel was investigated in rat hippocampal CA1 interneurons in slices using single-cell RT-PCR and patch-clamp recording techniques. We tested for the expression of the 5-HT_3A (both short and long splice variants) and 5-HT_3B subunits, as well as the expression of the α4 subunit of the neuronal nicotinic ACh receptors (nAChRs), the latter of which has been shown to co-assemble with the 5-HT_3A subunit in heterologous expression systems. Both the 5-HT_3A-short and α4-nAChR subunits were expressed in these interneurons, but we could not detect any expression of either the 5-HT_3B or the 5-HT_3A-long subunits. Furthermore, there was a strong tendency for the 5-HT_3A-short and α4-nAChR subunits to be co-expressed in individual interneurons. To assess whether there was any functional evidence for co-assembly between the 5-HT_3A-short and α4-nAChR subunits, we used the sulphydryl agent 2-aminoethyl methanethiosulphonate (MTSEA), which has previously been shown to modulate expressed 5-HT₃Rs that contain the α4-nAChR subunit. In half of the interneurons examined, MTSEA significantly enhanced the amplitude of the 5-HT₃R-mediated responses, which is consistent with the notion that the α4-nAChR subunit co-assembles with the 5-HT_3A subunit to form a native heteromeric 5-HT₃R channel in rat CA1 hippocampal interneurons in vivo. In addition, the single-channel properties of the 5-HT₃R were investigated in outside-out patches. No resolvable single-channel currents were observed. Using non-stationary fluctuation analysis, we obtained an estimate of the single-channel conductance of 4 pS, which is well below that expected for channels containing both the 5-HT_3A and 5-HT_3B subunits.

The superfamily of ligand-gated ion channels is a group of neurotransmitter receptors that form ion channels in the outer membranes of nerve cells. This family includes receptors for the excitatory neurotransmitters ACh (the neuronal nicotinic ACh receptors; nAChRs) and serotonin (the 5-HT₃ receptors; 5-HT₃Rs), and the inhibitory neurotransmitters γ-aminobutyric acid (both the GABA_A and GABA_C receptors) and glycine (GlyRs; for reviews see Betz, 1990; Jackson, 1999). These different receptor channels are not only functionally related, they are also structurally related. Each receptor channel complex is a pentameric assembly of individual protein subunits, where each subunit is thought to have four transmembrane domains, the second of which lines the ion channel pore (Betz, 1990; Jackson, 1999). Similar to other neurotransmitter receptor families, functional diversity is created by the assembly of different subunits; the assembly of different subunits can affect various properties of the ion channel, including the activation and desensitization kinetics, pharmacology and ionic permeability. Within the ligand-gated ion channel superfamily, there are 18 known GABA_A receptor subunit genes, 12 neuronal nAChR subunit genes, five GlyR subunit genes and two 5-HT₃R subunit genes.

The 5-HT₃Rs are distributed widely in the central and peripheral nervous systems, where they participate in a variety of physiological processes, including cognitive processing (Maeda et al. 1994; Staubli & Xu, 1995; Yakel, 2000). In the central nervous system, 5-HT₃Rs are expressed in the forebrain (e.g. cerebral cortex, hippocampus and amygdala), hindbrain, medulla oblongata, spinal cord and, to a lesser extent, in the nucleus accumbens, striatum and substantia nigra. Two 5-HT₃R subunit genes (i.e. 5-HT_3A and 5-HT_3B) have thus far been cloned (Maricq et al. 1991; Davies et al. 1999; Dubin et al. 1999; Hanna et al. 2000). For the 5-HT_3A subunit, a shorter splice variant (in which five or six amino acid residues located within the putative large intracellular loop between the third and fourth transmembrane domains are deleted) has also been cloned from mouse, rat and guinea pig (Hope et al. 1993; Miquel et al. 1995; Lankiewicz et al. 1998). The short form of the 5-HT_3A subunit is the most abundant form in the central nervous system (Miquel et al. 2002). Interestingly, humans possess only the short form of the 5-HT_3A subunit (Werner et al. 1994; Belelli et al. 1995; Miyake et al. 1995). In rats and in neuroblastoma cell lines, the relative proportion of the short and long forms of the 5-HT_3A subunit is developmentally regulated. In rats at embryonic day 17, the relative percentage of the long form increased from 10 % to 35 % in the hippocampus and cortex, and to 50-75 % in the superior cervical ganglion and nodose ganglia (Miquel et al. 1995). In NG108-15 neuroblastoma cells, the relative proportion of the two splice variants was also regulated in a similar fashion by differentiation (Emerit et al. 1995). The recently cloned 5-HT_3B subunit, which was found to be expressed in the hippocampus as well as other parts of the brain (Davies et al. 1999; Dubin et al. 1999; Monk et al. 2001), does not form channels on its own, but requires co-assembly with the 5-HT_3A subunit. The most notable alteration in functional property due to co-expression with the 5-HT_3B subunit was a significantly larger single-channel conductance (Davies et al. 1999; Hanna et al. 2000).

Although these multiple 5-HT₃R subunits and/or splice variants are known to exist, the diverse functional and pharmacological properties of native 5-HT₃Rs cannot be fully accounted for by these known subunits, suggesting the possibility that other as yet unknown 5-HT₃R subunits exist that might generate functional diversity in vivo (Fletcher & Barnes, 1998; Yakel, 2000). While there still may be as yet uncloned 5-HT₃ subunits, there is in vitro evidence that other subunits might be participating. For example, Hussy et al. (1994) suggested that nAChR subunits, or subunits from another ligand-gated ion channel, might associate with 5-HT₃R subunits. The 5-HT₃Rs purified from porcine brain contain both 5-HT_3A and non-5-HT_3A subunit proteins (Fletcher & Barnes, 1997), and the 5-HT₃R and α4-nAChR subunits colocalize on a subset of rat striatal and cerebellar synaptosomes (Nayak et al. 2000). Recently, it has been reported that the α4-nAChR subunit can co-assemble with the 5-HT₃R subunit in human embryonic kidney (HEK)-293 cells and Xenopus oocytes to form a functional channel with altered functional properties (van Hooft et al. 1998; Kriegler et al. 1999a; Harkness & Millar, 2001). However, it should be noted that Fletcher & Barnes (1998) recently reported that the non-5-HT_3A subunit proteins purified from porcine brain along with the 5-HT₃R were neither the α1-, α3-, α4, α5-, α7- nor the β2-nAChR subunits. Nevertheless, whereas co-assembly has been demonstrated in vitro, whether or not such co-assembly exists in vivo, and which subunits might be assembling with the 5-HT₃Rs, remains to be determined.

To investigate the molecular makeup of 5-HT₃Rs in the rat hippocampus, we examined the 5-HT₃R-gated whole-cell and single-channel currents and the subunit expression patterns of the 5-HT_3A (both short and long splice variants), 5-HT_3B, and α4-nAChR subunits in rat CA1 hippocampal interneurons. This population of neurons has previously been shown to functionally express both nAChRs (Jones & Yakel, 1997; Alkondon et al. 1998; Frazier et al. 1998; McQuiston & Madison, 1999; Sudweeks & Yakel, 2000) and 5-HT₃Rs (McMahon & Kauer, 1997). However, the molecular composition of the 5-HT₃Rs expressed by these neurons is unknown.

In the present study we used patch-clamp recording techniques to characterize ion channel function, and single-cell RT-PCR techniques to analyse subunit mRNA expression. There is no evidence, either functional or molecular, for the heteromeric co-assembly of the 5-HT_3A and 5-HT_3B subunits in these hippocampal interneurons. In addition, we found no evidence to suggest that the 5-HT_3B nor the 5_-HT_3A-long subunits are expressed in these neurons. Instead we provide evidence, at least in some cells, for the co-expression and co-assembly of the 5-HT_3A-short and α4-nAChR subunits: the first demonstration of co-assembly of subunits from diverse ligand-gated ion channels in vivo.

Methods

Hippocampal slice preparation

All experiments were carried out in accordance with guidelines approved by the NIEHS Animal Care and Use Committee. Wistar rats (11-19 days old) were anaesthetized with halothane and then decapitated. The brain was removed and immediately placed into ice-cold artificial cerebrospinal fluid (ACSF) bubbled with 95 % O₂-5 % CO₂. ACSF consisted of (mm): 126 NaCl, 3.5 KCl, 1.2 NaH₂PO₄, 25 NaHCO₃, 11 glucose, 1.3 MgCl₂, 2 CaCl₂, pH 7.4. Coronal slices (300 μm thick) were cut using a vibratome (Series 1000, Ted Pella, Redding, CA, USA), and parasaggital slices (250 μm thick) were cut using a vibroslicer (752 M, Campden Instruments, UK; van Hooft, 2002). Slices were then transferred to a chamber containing ACSF and bubbled with 95 % O₂-5 % CO₂ at room temperature for at least 1 h before recording.

Patch-clamp electrophysiology

Hippocampal slices were held in the recording chamber by using a platinum bridge with nylon cross fibres. During experiments, the slices were perfused with ACSF (room temperature, bubbled with 95 % O₂-5 % CO₂) at 1-2 ml min⁻¹. Functional 5-HT₃R currents were isolated pharmacologically by including atropine (10 μm, to block muscarinic ACh receptors), tetrodotoxin (TTX, 1 μm, to block Na⁺ channels), 6-cyano-7-nitroquinoxaline-2, 3-dione (10 μm, to block glutamate receptors), 2-amino-5-phosphonovaleric acid (10 μm, to block NMDA receptors), and picrotoxin (50 μm, to block GABA_A receptors) in the perfusing ACSF (normally K⁺ channels were blocked by including Cs⁺ in the internal solution; see below). All drugs were obtained from Sigma. Currents elicited from hippocampal interneurons by the application of serotonin (5-HT) were identified pharmacologically as 5-HT₃R-mediated currents by their sensitivity to blockade by MDL-72222 (Sigma-RBI).

Neurons were identified visually, using differential interference contrast (DIC) optics, by location in the slice, in the CA1 stratum radiatum, stratum lacunosum moleculare and stratum oriens layers of the hippocampus. Recording/aspiration electrodes were pulled from Corning 7052 filamented glass (Garner Glass, Claremont CA, USA) using a P-97 micropipette puller (Sutter Instruments, Novato, CA, USA) to a tip diameter of approximately 3 μm. The intrapipette solution consisted of (mm): 140 caesium gluconate or CsCl, 2 MgCl₂, 0.5 CaCl₂, 5 BAPTA or EGTA, 10 Hepes, and 2 Mg-ATP.

Neurons were voltage clamped at -70 mV during recordings (unless otherwise specified; corrected for a 10 mV junction potential when using caesium gluconate). Whole-cell patch-clamp recordings were obtained using either an Axopatch 200A amplifier (Axon Instruments, Foster City, CA, USA) or an EPC9 patch-clamp amplifier (HEKA Electronic, Lambrecht, Germany). Whole-cell recordings were filtered at 1-5 kHz, and sampled at 2-10 kHz, using either PCLAMP Clampex software (version 8, Axon Instruments), or PULSE software (HEKA Electronic). 5-HT was applied by pressure ejection for 100 or 500 ms at 35-140 kPa pressure using a picospritzer II (General Valve, Fairfield, NJ, USA) connected to a glass pipette placed 60-100 μm from the cell body. A washout period of at least 3 min was included between subsequent recordings to avoid receptor desensitization. For studying the properties of the action potentials, neurons were recorded from under current-clamp conditions, utilizing potassium gluconate instead of caesium gluconate for the intrapipette solution. Neurons were defined as fast-spiking if the rate of evoked action potential firing was greater than 40 Hz.

Non-stationary fluctuation analysis

Outside-out recordings were made using an EPC9 patch-clamp amplifier and PULSE software (HEKA Electronic). Patch pipettes were pulled from boroscilicate glass and had a resistance of 3-5 MΩ when filled with internal solution, which consisted of (mm): 140 caesium gluconate, 0.5 CaCl₂, 5 EGTA, 10 Hepes and 2 Mg-ATP, pH 7.3 with CsOH. Cells were voltage clamped at -60 mV. A second pipette, connected to a picospritzer (General Valve), was positioned in the vicinity of the cell soma and 5-HT (100 μm) was applied for 100 ms at 35-105 kPa pressure. Outside-out patches were made from cells that responded with a fast inward current of at least 50 pA in the whole-cell configuration. The outside-out patch was positioned in front of the opening of the pipette connected to the picospritzer, and 5-HT was applied with an interval of 60 s. Signals were filtered at 10 kHz and sampled at 20 kHz.

Non-stationary fluctuation analysis was performed as described previously (Jonas et al. 1993). Six to ten responses per outside-out patch were averaged, and the variance around the mean was calculated. The relationship between the mean current amplitude (I) and the variance (σ²) was fitted with the equation σ² = iI - I²/N, where i is the single-channel current amplitude and N is the estimate of the number of channels open at the peak of the current. The maximal open probability of the channels (P_o,max) was subsequently calculated as: P_o,max = 1 - (σ_peak²/iI_peak), where σ_peak² and I_peak are the variance and the amplitude of the current at its peak, respectively. Off-line analysis was performed using IGOR Pro (Wavemetrics, OR, USA). All data are given as mean ± s.e.m. of n independent experiments.

Oocyte recordings

mRNA was transcribed in vitro from plasmids using mMessage Machine kit (Ambion, Austin, TX, USA) according to conditions suggested by the manufacturer; plasmid DNA was kindly provided by Dr James Patrick (Baylor College of Medicine, Houston, TX, USA; rat α4-nAChR), and Dr David Julius (UCSF, San Francisco, CA, USA; mouse 5-HT_3A-long). All experiments were carried out in accordance with guidelines approved by the NIEHS Animal Care and Use Committee. Female Xenopus laevis frogs were anaesthetized by immersion in cold water containing 0.2 % ethylmetaaminobenzoate (MS-222; Sigma) for 60 min, and then decapitated. Oocytes were dissected and defolliculated by treatment with collagenase B (Boehringer Mannheim, 3-4 mg ml⁻¹) for 2-4 h in a solution containing (mm): 85.2 NaCl, 2 KCl, 1 MgCl₂, and 5 Hepes (pH 7.4; Kriegler et al. 1999a). The total amount of RNA injected for the 5-HT_3A subunit was 0.5 ng, and for the α4-nAChR subunit was 25 ng. Experiments were performed 2-4 days after injection.

Current responses were obtained by two-electrode voltage-clamp recording at a holding potential of -60 mV using a Geneclamp 500 system and pCLAMP 8 software (Axon Instruments). Typically, traces were filtered at 0.2 kHz and sampled at 0.5 kHz. Electrodes contained 3 m KCl with 0.4 m BAPTA and had a resistance of < 1 MΩ. 5-HT was freshly prepared in bath solution from a frozen stock and applied via a synthetic quartz perfusion tube (0.7 mm i.d.) operated by a computer-controlled valve. Recordings were performed in a solution containing (mm): 96 NaCl, 2 KCl, 1.8 CaCl₂, 1 MgCl₂, and 10 Hepes (pH 7.4). Oocytes were maintained in culture in the same solution, and with the addition of 2.5 mm sodium pyruvate, 0.5 mm theophylline and 50 μg ml⁻¹ gentamicin.

2-Aminoethyl methanethiosulphonate (MTSEA) application

The positively charged MTSEA compound (Toronto Research Chemicals, Ontario, Canada) is a sulphydryl agent that reacts with cysteine residues (Akabas et al. 1994; Karlin & Akabas, 1998). After obtaining several stable amplitude 5-HT responses (i.e. 2-5 cell⁻¹), MTSEA (1 mm) was mixed freshly into either ACSF that had been pre-bubbled with 95 % O₂-5 % CO₂ or oocyte bath solution, and immediately applied to the cells for 4 min. Another 5-HT-gated response was then recorded during the continued presence of MTSEA.

Reverse transcription reaction

Interneurons from acutely prepared hippocampal slices were aspirated under visual observation by application of suction using a 10 cc syringe attached to the aspiration pipette (same dimensions as a recording pipette), and then added to the reverse transcription (RT) reaction mixture. The RT mix consisted of (mm, unless stated otherwise): 50 KCl, 10 Tris-HCl (pH 8.3), 5 MgCl₂, 1 DTT, 1 each dNTP, 2.5 μm random hexamers, 0.33 U μl⁻¹ RNase inhibitor (5 Prime-3 Prime, Boulder, CO, USA), RNA and 10 U μl⁻¹ Superscript II reverse transcriptase (Gibco-BRL, Rockville, MD, USA) in diethyl pyrocarbonate (DEPC)-treated water. RT reactions had a total volume of 10 μl each. Reactions were run at 25 °C for 10 min, 42 °C for 60 min and 95 °C for 5 min in a PTC-200 thermal cycler (MJ Research, Watertown, MA, USA). Reactions were stored at 4 °C until running the PCR.

Polymerase chain reaction

The first round of multiplex PCRs was performed by adding the following to the completed RT reaction (mm, unless stated otherwise): 50 KCl, 10 Tris-HCl (pH 8.3), 2.5 MgCl₂, 0.25 each dNTP, 0.2 μm each upstream and downstream primer for the α4-nAChR subunit, the 5-HT_3A subunit and the 5-HT_3B subunit, along with 0.1 μm of each β-actin primer, and 0.05 U μl⁻¹ Platinum Taq polymerase (Gibco-BRL) in DEPC-treated water to a final volume of 25 μl. The following primers are listed as sequence(s) amplified, upstream primer sequence, downstream primer sequence, size of product: α4 (also amplifies the nAChR α7 subunit), ATTGATGTGGATGAGAAGAACCA, AGCAGGAAGACGGTGAGAGAAAG, 650 base pairs (bp); 5-HT_3A, CTGACTGGCTGAGGCACCTG, GGTGGTGGAAGAGGGCTATC, 205/220 bp (short/long splice variants); 5-HT_3B, CGGCACAGTGAGCAAGAACG, AGTGGCACAGGATGGCGAGC, 244 bp; β-actin, AAGATCCTGACCGAGCGTGG, CAGCACTGTGTTGGCATAGAGG, 327 bp. All primers were made by Life Technologies (Gibco-BRL). PCR was held at 94 °C for 30 s then cycled 25 times. Each cycle consisted of: 92 °C for 15 s 59 °C for 20 s and 72 °C for 30 s. Two-microlitre samples of the initial multiplex PCR were used as substrates for each sequence-specific reaction in the second round of PCRs. A 25 μl reaction sample was prepared for each sequence analysed. These reactions consisted of (mm, unless otherwise stated): 50 KCl, 10 Tris-HCl (pH 8.3), 1.5-4.0 MgCl₂, 0.25 each dNTP, 1.0 μm upstream- and downstream-specific primers, and 0.025 U μl⁻¹ Platinum Taq polymerase in DEPC-treated water. The second-round primers for the nAChR α4 sequence were CCAGATGATGACAACCAACG, CCACACGGCTATGAATGCTC (nested product size = 356 bp); all other primer sequences were the same as in the first round.

Reactions were held at 92 °C for 30 s then cycled 40 times as follows: 92 °C for 15 s annealing temperature (55-63 °C) for 20 s and 72 °C for 30 s; 7.0 μl samples were analysed on 2 % agarose gels using ethidium bromide. The presence or absence of a subunit was scored by whether or not a clearly defined band of the correct size was detectable by visual observation of the photographed gel. Non-specific bands occasionally appeared, as would be expected when running 40 PCR cycles, after 25 cycles of multiplex PCR with several primer pairs. Spurious bands of the incorrect sizes were ignored, and DNA bands from several experiments were sequenced to verify subunit specificity of the RT-PCR protocol.

RT-PCR controls

Perfusion solutions were made from DEPC-treated water and vacuum-filtered using 0.2 μm cellulose nitrate filters (Nalge Nunc, Rochester, NY, USA). Solutions were tested periodically using the RNAse Alert system to ensure that they had no inherent RNase activity (Ambion). Gloves were worn during all experiments and recordings to minimize the chance of contaminating samples with RNases.

Primers were designed using the Vector NTI software package (Informax, Bethesda, MD, USA). Primer optimization and sensitivity was tested for the α4-nAChR, 5-HT_3A and 5-HT_3B subunits using both rat brain RNA and plasmid DNA; plasmid DNA was kindly provided by Dr James Patrick (rat α4-nAChR), Dr David Julius (mouse 5-HT_3A-long) and Dr Ewen Kirkness (rat 5-HT_3B) of the Institute for Genomic Research (Rockville, MD, USA). The sensitivity of the PCR primers and the optimized amplification protocols was tested by amplifying from 10:1 serial dilutions of plasmid stock. Using this procedure, the amplification limit (i.e. the lowest DNA concentration from which successful PCR amplification reliably occurred), measured using the two-step PCR protocol outlined above, was 0.1 fg for the nAChR α4 subunit and 10 fg for both the 5-HT_3A and 5-HT_3B subunits. Since the detection sensitivity was the same for both 5-HT₃ subunits, a direct comparison of the results for RT-PCR subunit detection of these two subunits is possible.

An extract of mRNA expressed in the brain (whole brain minus the olfactory bulb and cerebellum) of an 18-day-old rat was prepared by immediately homogenizing an acutely dissected brain in Tri-reagent (Sigma, using 1 ml (100 mg tissue)⁻¹) and extracting the RNA fraction following the manufacturer's protocol. To test the sensitivity of the RT-PCR protocol for the 5-HT_3A and 5-HT_3B receptor subunits on actual rat brain RNA, 10:1 serial dilutions (starting at 1 μg μl⁻¹) were prepared in DEPC-treated water, and then used as the substrate for RT-PCR. These reactions showed that 5-HT_3A could be detected down to a dilution of 0.1 ng RNA, while the 5-HT_3B subunit could only be detected down to a dilution of 10 ng RNA, indicating that the 5-HT_3A subunit is expressed at much higher concentrations (approximately × 100) in the 18-day-old rat brain than the 5-HT_3B subunit. This data is consistent with our single-cell RT-PCR data from hippocampal interneurons (see Results).

Negative RT-PCR controls were performed in conjunction with each set of neurons by aspirating ACSF directly from the surface of the hippocampal slice in the recording chamber and analysing it in parallel with the aspirated neurons for all mRNAs examined in that set. The false-positive rate was calculated by taking the total number of positive PCR reactions (2 for β-actin) from the negative control sets and dividing by the total number of negative control reactions performed (53). The overall false-positive detection rate was 3.8 % for our RT-PCR protocol. No false-positive PCR reactions for either the 5-HT_3A, 5-HT_3B or α4-nAChR receptor subunits were ever detected. β-Actin RT-PCR detection was used as a positive control for cell aspiration, since it is highly expressed by all neurons. Our overall detection of β-actin in these interneurons was 92.5 %, indicating a high rate of successful cell aspiration.

Statistical analysis

For the RT-PCR subunit detection analysis, the proportion of neurons expressing a particular mRNA is equal to the number of observed positive neurons divided by the total number of neurons analysed for that mRNA transcript. The standard error of the proportion (s.e.p.) was calculated for each mRNA transcript, and a z-test for comparing two proportions was used as described previously (Sudweeks & Yakel, 2000). Pearson's correlation coefficients (r) were calculated (using the Excel PEARSON function) for comparing mRNA co-expression, and their statistical significance was tested using procedures also detailed previously (Sudweeks & Yakel, 2000). R values can range from +1 to -1, where a value of +1 signifies an exact positive correlation. An r value of zero would be expected from a completely random association.

Results

Single-cell RT-PCR to characterize ion channel subunit expression

The whole-cell patch-clamp recording technique was utilized to examine 5-HT₃R-mediated responses in rat hippocampal CA1 interneurons in the slice. The application of 5-HT activated an inward current response (at a holding potential of -60 mV) in 28 % of the 178 interneurons examined. It was determined that this current was due to the activation of the 5-HT₃R because it was blocked by the selective antagonist MDL-72222 (Fig. 1A). To characterize the subunit composition of these 5-HT₃Rs, we used single-cell RT-PCR techniques (Sudweeks & Yakel, 2000) to analyse the expression of the 5-HT_3A (both short, 5-HT_3A-short, and long, 5-HT_3A-long, forms), 5-HT_3B and α4-nAChR subunit mRNAs in 50 hippocampal interneurons in the CA1 stratum radiatum, stratum lacunosum moleculare and stratum oriens layers. Overall, we successfully detected single-cell expression of the 5-HT_3A-short (but not the 5-HT_3A-long) subunit in 18 ± 5 % of interneurons examined (mean ± s.e.p.; see Methods), and the α4-nAChR subunit in 20 ± 6 % of these neurons (Fig. 1B and C). We did not detect any expression of the 5-HT_3B subunit in individual interneurons, although we could detect this subunit in a whole brain (minus the olfactory bulb and cerebellum) RNA preparation (see Methods). To determine whether our RT-PCR assay was as sensitive for the detection of the 5-HT_3B subunit as the 5-HT_3A subunit, we performed a sensitivity assay using plasmid DNA for these subunits (see Methods). This assay indicated that our PCR protocol was equally sensitive in detecting the 5-HT_3A and 5-HT_3B subunits. Thus we can conclude from our data that there is a significantly higher expression of the 5-HT_3A-short than for the 5-HT_3B subunit in these hippocampal interneurons. We cannot rule out the possibility of 5-HT_3B mRNA expression in these cells at a level below our RT-PCR detection threshold.

Figure 1. 5-HT₃ receptor (5-HT₃R)-mediated currents and single-cell RT-PCR analysis in rat CA1 hippocampal interneurons.

A, the pressure application of 5-HT (100 μm for 500 ms; see Methods) induced an inward current response (at a holding potential of -60 mV) that was blocked by the bath application of the specific 5-HT₃R antagonist, MDL-72222 (100 nm). B, example agarose gels of the single-cell RT-PCR products. The lanes marked α4 and 5-HT_3A are from the same individual hippocampal interneuron (located at the strata radiatum/lacunosum border) and indicate the co-expression of the α4 subunit of the neuronal nicotinic ACh receptor (α4-nAChR) and 5-HT_3A subunit mRNA from this cell. Marker lanes (M) show a 100 base pairs (bp) ladder, and the expected band sizes for the α4 (356 bp) and 5-HT_3A (205 bp) subunits are indicated with arrows. The fainter band apparent on the α4 gel at 205 bp is most likely a carryover PCR product from the 5-HT_3A primers that was also present in the initial 25 cycles of multiplex PCR. The middle lanes (two on the left gel for the α4-nAChR subunit mRNA and one on the right gel for 5-HT_3A subunit mRNA) are from different interneurons where the PCR product was not detected. C, the proportion of individual cells where we detected each particular mRNA examined. The experimentally determined detection limits are reported in Methods. Results were obtained from 50 individual interneurons.

We investigated whether there was a tendency for the 5-HT_3A and α4-nAChR subunits to be co-expressed in individual interneurons. Using the individual RT-PCR detection rates for these two subunits calculated above (18 % and 20 %, respectively), only 3.6 % (i.e. 18 % × 20 %) of the cells examined would be expected to express both subunits if these two subunits were expressed independently. However, in the interneurons examined above, the actual percentage of cells expressing both was 8 %. Furthermore, of the 5-HT_3A subunit-expressing cells, 44 % also had detectable levels of the α4-nAChR subunit, as compared to only 20 % of the general population of cells expressing the α4-nAChR subunit. Of the α4-nAChR subunit-expressing cells, 44 % also had detectable levels of the 5-HT_3A subunit (as compared to 18 % in the general population). Pearson's correlation coefficient (see Methods) for co-expression of the 5-HT_3A and α4-nAChR subunits was 0.4 (if randomly distributed, this value would be 0), and the P value (indicating whether this coefficient is significantly different from 0) is 0.006 (see Methods). Interestingly, this correlation coefficient is even greater than the correlation coefficient we obtained previously for the α4 subunit with the β2 nAChR subunit (r = 0.19; Sudweeks & Yakel, 2000), which forms the major high-affinity nicotine binding site in the brain (the α4β2-nAChRs; Jones & Yakel, 1997). These data indicate that there is a tendency for the 5-HT_3A-short and the α4-nAChR subunits to be co-expressed in individual interneurons.

Initially, we performed the single-cell RT-PCR analysis in conjunction with whole-cell patch-clamp recordings (25 cells) in order to correlate function with expression pattern, as we had done previously for various nAChR subunits in these interneurons (Sudweeks & Yakel, 2000). However, for the present study, the experimental conditions were different in that a much longer period of time elapsed between obtaining the whole-cell configuration and the aspiration of the cells for the single-cell RT-PCR analysis. Several baseline recordings of stable 5-HT-gated responses (average of four per cell) were needed from each cell studied, with a washout period of at least 3 min per application, to test for the effects of MTSEA (see below); thus an extended period of time elapsed for each successful recording. This elapsed time decreased our successful RT-PCR detection rates from these cells for the α4-nAChR subunit to 0 % (as compared to 20 % when recordings were not done), and for the positive control β-actin to 83 % (as compared to 100 %). Thus, because of our inability to detect the α4-nAChR subunit after recordings in the present study, we could not correlate the function of the 5-HT-gated responses with the presence or absence of the α4-nAChR subunit.

Interestingly, there was a slight increase in the cells where we detected the 5-HT_3A subunit (24 % as compared to 18 %) after recordings. This allowed us to make a comparison between the cells where we detected the 5-HT_3A subunit and functional properties, in particular the amplitude of the 5-HT-gated responses as an indicator of protein expression levels (Fig. 2). In the cells where we detected the 5-HT_3A subunit, the average peak amplitude was 77 ± 30 pA (6 cells), whereas in cells where we did not detect this subunit, the average amplitude was 19 ± 5 pA (19 cells); for 7 of these latter cells, there was no 5-HT-induced current (Fig. 2). Thus, there was a correlation between the protein levels (as measured by peak current amplitude) and the RT-PCR detection of mRNA for the 5-HT_3A subunit.

Figure 2. Correlation between the amplitude of 5-HT₃R-mediated currents and the detection of 5-HT_3A subunit mRNA.

A, two superimposed recordings from interneurons, one with a large 5-HT-induced response, and the other with a barely detectable response, in response to the application of 5-HT (100 μm for 100 ms). B, the gel shows four lanes plus the marker. The two outside lanes (1 and 4) correspond to the cells recorded from (marked by stars), and the expected band size for the 5-HT_3A subunit is shown by the arrows (205 bp). Marker lanes (M) show a 100 bp ladder. Lane 2 is another interneuron (which was negative for the 5-HT_3A subunit) and lane 3 is a negative control for the 5-HT_3A subunit. All three neurons were positive for β-actin, and the control was negative.

Functional co-assembly: MTSEA modulation of 5-HT₃R currents

The expression pattern of the 5-HT_3A-short and α4-nAChR subunits were strongly correlated in individual neurons, and previously these two subunits were found to co-assemble to form a functional heteromeric channel in heterologous expression systems (van Hooft et al. 1998; Kriegler et al. 1999a). As such, we were interested in testing whether these two diverse receptor subunits might be co-assembling to form a functional channel in rat hippocampal interneurons. To test this, we used the cysteine-reactive compound MTSEA, which has been shown previously to modulate expressed neuronal nAChRs containing the α4 subunit (Yu et al. 1996), as well as expressed 5-HT₃Rs containing the α4-nAChR subunit (Kriegler et al. 1999a,b). When homomeric 5-HT_3ARs were expressed in Xenopus oocytes, the application of MTSEA (1 mm) did not affect the amplitude of the 5-HT-gated current (Fig. 3A). However, when the α4-nAChR subunit was co-expressed with the 5-HT_3A subunit, the 5-HT-gated currents were now sensitive to modulation by MTSEA (Fig. 3B); the average peak current increased by 14 ± 0.5 % (four cells), indicating the presence of the α4-nAChR subunit assembled with the 5-HT-gated channel (Kriegler et al. 1999b).

Figure 3. 2-Aminoethyl methanethiosulphonate (MTSEA) potentiates 5-HT₃R-mediated responses in Xenopus oocytes only when co-expressed with the α4-nAChR subunit.

The bath application of MTSEA (1 mm for 4 min; see Methods) had no effect on 5-HT-gated currents in oocytes expressing homomeric 5-HT_3ARs (A), but it potentiated currents (by 15 % for this particular oocyte) in oocytes where the α4-nAChR subunit was co-expressed with the 5-HT_3AR subunit (B).

As MTSEA was an effective tool for investigating co-assembly of the 5-HT_3A and α4-nAChR subunits in heterologous expression systems, we tested whether MTSEA modulated 5-HT₃R-mediated currents in hippocampal interneurons. This would suggest the possibility that these two subunits are co-assembling in vivo. The application of MTSEA (1 mm), which had no effect on the baseline current level, significantly increased the amplitude of the 5-HT-gated currents in eight of 16 cells (by an average of 52 ± 16 %; Fig. 4); MTSEA either had no affect or caused a modest reduction in amplitude in the remaining eight cells (the average decrease in these cells was 7 ± 4 %). The fact that half of the 5-HT-gated currents were modulated by MTSEA is consistent with our single-cell RT-PCR data, indicating that some 5-HT₃Rs also contain the α4-nAChR subunit.

Figure 4. MTSEA potentiates 5-HT₃R-mediated responses in rat hippocampal interneurons.

The bath application of MTSEA (1 mm for 4 min) potentiated the amplitude of the 5-HT-gated current, by 97 % for this particular neuron. The cell was held at a potential of -70 mV, and the concentration of 5-HT was 100 μm.

Estimation of single-channel conductance using noise analysis

It was shown previously that co-assembly of the 5-HT_3B subunit with the 5-HT_3A subunit dramatically increases the single-channel conductance of the 5-HT₃R channels. When expressed in HEK-293 cells, no resolvable single-channel currents were observed with the 5-HT_3A subunit alone, but single-channel events were observed when the 5-HT_3A and 5-HT_3B subunits were co-expressed (Davies et al. 1999; Hanna et al. 2000). As we failed to detect the 5-HT_3B subunit mRNA in rat hippocampal interneurons, we wanted to investigate the single-channel properties of the 5-HT₃R channels in order to determine whether these data were consistent with the absence of the 5-HT_3B subunit. In excised outside-out patches (six patches), we were unable to observe any resolvable single-channel currents (Fig. 5A). To obtain an estimate of the single-channel conductance of the 5-HT₃R channels in these interneurons, non-stationary fluctuation analysis was performed (Jonas et al. 1993). Application of 100 μm 5-HT to outside-out patches from the soma of stratum radiatum interneurons induced a small inward current of 1.3 ± 0.6 pA (six patches). The fits of the variance-mean plot (Fig. 5B) gave a single-channel amplitude of 247 ± 110 fA at a holding potential of -60 mV. Assuming that the reversal potential of the 5-HT₃R-mediated ion current is around 0 mV, this results in an estimate of the single-channel conductance of 4.3 ± 1.4 pS, well below that expected for channels containing both the 5-HT_3A and 5-HT_3B subunits. The number of channels open at the peak of the current (N) and P_o,max were estimated to be 20 ± 16 and 0.19 ± 0.07, respectively.

Figure 5. Non-stationary fluctuation analysis of 5-HT₃R-mediated ion currents.

A, six superimposed 5-HT₃R-mediated ion currents (grey traces) and the mean of those currents (black trace), recorded from an outside-out patch. B, plot of the variance against the mean of the current. The continuous line represents the fit according to σ² = iI - I²/N (see Methods). γ, single channel conductance.

Action potential properties of 5-HTHT₃R-containing interneurons

We have investigated the action potential firing properties of interneurons using a potassium-based intrapipette solution and without TTX in the bath solution. Of 11 stratum radiatum interneurons with 5-HT₃R-mediated responses, eight had a regular-spiking action potential firing profile, whereas the rest showed fast-spiking properties (Fig. 6). This is in line with the reported heterogeneous spiking properties of cholecystokinin (CCK)-containing interneurons in the CA1 region of the rat hippocampus (Pawelzik et al. 2002), about ≈60 % of which were previously shown to express the 5-HT₃R (Morales & Bloom, 1997).

Figure 6. Heterogeneous spiking properties of 5-HT₃R-expressing interneurons.

Examples of responses from a regular-spiking (A) and fast-spiking (B) interneuron. Both interneurons were located in the stratum radiatum and responded to 5-HT application with 5-HT₃R-mediated inward current responses. Action potentials were evoked by current injection (+150 pA and -100 pA for the upper and lower traces, respectively), and the resting membrane potentials were -71 mV (A) and -68 mV (B). The averaged action potential firing rate was 26 ± 5 Hz for regular-spiking interneurons (n = 8) and 58 ± 11 Hz for fast-spiking interneurons (n = 3).

Discussion

We have shown, using single-cell RT-PCR and whole-cell patch-clamp recording techniques, that the 5-HT_3A-short and α4-nAChR subunits are co-expressed in individual rat hippocampal CA1 interneurons. We found no evidence for the presence of either the 5-HT_3B subunit or the 5HT_3A-long subunits. In addition, we were unable to observe discrete single-channel current events, as would be expected for 5-HT₃Rs assembled from both 5-HT_3A and 5-HT_3B subunits. These molecular and functional data are not consistent with the expression of the 5-HT_3B subunit in rat hippocampal interneurons. Therefore, we conclude that the 5-HT_3A-short subunit is most likely the primary 5-HT₃R subunit contributing to the formation of functional 5-HT₃Rs in rat hippocampal interneurons. Furthermore, our data suggest the possibility that the 5-HT_3A and α4-nAChR subunits, which have previously been found to co-assemble in heterologous expression systems (van Hooft et al. 1998; Kriegler et al. 1999a), co-assemble in these neurons in vivo. The sulphydryl agent MTSEA potentiated 5-HT₃R-mediated responses in half of the interneurons examined. This functional data is consistent with the single-cell RT-PCR data (showing that 44 % of 5-HT_3A subunit-expressing cells also had detectable levels of the α4-nAChR) and the idea of co-assembly between the 5-HT_3A and α4-nAChRs subunits in these neurons (Kriegler et al. 1999a). This also is consistent with the notion of heterogeneity in receptor subunit composition; i.e. that some 5-HT₃Rs may be homomeric, while others may be heteromeric and include both 5-HT_3A and α4-nAChR subunits. The co-assembly of subunits from diverse neurotransmitter receptors provides a possible mechanism for creating functional diversity among neurotransmitter receptors without the requirement of having extra subunit genes. Co-assembly between the 5-HT_3A and α4-nAChRs subunits may not be surprising when considering the degree of similarity between them.

Prior to the cloning of the 5-HT_3B subunit, it was known that the properties of native 5-HT₃R-mediated responses could not be accounted for simply by expression of the short and/or long forms of the 5-HT_3A subunit (Fletcher & Barnes, 1998; van Hooft & Vijverberg, 2000). One physiological characteristic that often differentiated native 5-HT₃R channels from those expressed in heterologous expression systems was the single-channel conductance values. Whereas observable single-channel currents (ranging from 6 to 27 pS) were often observed in native tissues, whether from neurons or in some cases from neuroblastoma cell lines (Shao et al. 1991; van Hooft & Vijverberg, 1995; Fletcher & Barnes, 1998), no observable single-channel conductances were observed from homomeric 5-HT₃Rs expressed in heterologous expression systems from the 5-HT_3A subunit alone; noise analysis estimates concluded that the conductance levels were subpicosiemens (Hussy et al. 1994; Brown et al. 1998). The cloning of the 5-HT_3B subunit appeared to solve this discrepancy because when this subunit was co-expressed with the 5-HT_3A subunit, observable single-channel currents with conductances of 7-16 pS were now observable (Davies et al. 1999; Hanna et al. 2000).

In cultured hippocampal neurons, 5-HT₃R-mediated single-channel currents of ≈8-11 pS were observed (Jones & Surprenant, 1994). Furthermore, both the 5-HT_3A and 5-HT_3B subunits were reported to be expressed in the hippocampus, although it was not directly tested whether both subunits were co-expressed in individual neurons within the hippocampus (Davies et al. 1999; Dubin et al. 1999; Monk et al. 2001). However, using single-cell RT-PCR analysis and recordings from excised patches to study the single-channel currents, we did not observe single-channel currents nor find any other evidence for the expression of the 5-HT_3B subunit in individual hippocampal interneurons. To date, no functional evidence obtained either from the hippocampus or anywhere else in the nervous system has demonstrated that the 5-HT_3B subunit is participating in the formation of functional native 5-HT₃Rs. Therefore, the molecular makeup of the functional 5-HT₃Rs in rat hippocampal interneurons remains unresolved, and some other explanation must account for the diversity of the 5-HT₃Rs in these neurons. Our data suggest the possibility that co-expression with the α4-nAChR subunit is a possible explanation. It should be noted that we do not yet know whether co-expression with the α4-nAChR subunit results in observable 5-HT₃R-mediated single-channel currents. It should also be noted that while the 5-HT₃Rs purified from pig cerebral cortex contain both 5-HT_3A and non-5-HT_3A subunit proteins, Fletcher & Barnes (1997) did not detect the α4-nAChR subunit. We have suggested previously (van Hooft et al. 1998) that this apparent discrepancy is due to the fact that co-assembly between the 5-HT_3A and α4-nAChRs subunits might occur in other regions of the brain or in small fractions of specific cell populations. Therefore, perhaps biochemical investigations on pure populations of rat hippocampal interneurons might yield data consistent with co-assembly.

Hippocampal inhibitory interneurons form an extremely diverse group of heterogeneous cells. Utilizing a broad array of techniques and criteria, Parra et al. (1998) distinguished at least 16 different morphological types and at least three different modes of discharge. McMahon & Kauer (1997), who were the first to report on the functional properties of 5-HT₃R-mediated responses in rat hippocampal CA1 interneurons in slices, reported that the axons of these interneurons ramify widely within the CA1 region, and some also project to and arborize extensively in the dentate gyrus. Furthermore, they have shown clearly that the 5-HT₃R-expressing interneurons are heterogeneous with respect to anatomy and projections. Using in situ hybridization and immunocytochemistry, Morales & Bloom (1997) reported that 5-HT₃Rs were expressed in CCK-containing interneurons in the hippocampus; about 60 % of these interneurons expressed the 5-HT₃R. We have found that most (i.e. 73 %) of these functional 5-HT₃R-containing cells had a regular spiking action potential firing profile, whereas the rest showed fast-spiking properties. This is consistent with the reported heterogeneous spiking properties of CCK-containing interneurons in the CA1 region of the rat hippocampus (Pawelzik et al. 2002).

The Ca²⁺ permeability of 5-HT₃Rs was reported to be altered by co-expression with either the 5-HT_3B subunit, which lowered the Ca²⁺ permeability (Davies et al. 1999), or the α4-nAChR subunit, which raised the Ca²⁺ permeability (van Hooft et al. 1998). Although the relative Ca²⁺ permeability of the 5-HT₃Rs in rat hippocampal interneurons is not known, the function of the 5-HT₃Rs in these and other tissues is regulated by various signal transduction cascades (Yakel & Jackson, 1988; Yakel et al. 1991). Recently, Jones & Yakel (1998) have shown that the influx of Ca²⁺ through voltage-gated Ca²⁺ channels can modulate the rate of 5-HT₃R channel desensitization. Both protein kinase C (PKC) and the Ca²⁺-calmodulin-regulated protein phosphatase, calcineurin (van Hooft & Vijverberg, 1995; Zhang et al. 1995; Boddeke et al. 1996), also have been shown to regulate the function of the 5-HT₃R. It is important to determine whether or not the 5-HT₃Rs in the hippocampus are permeable to Ca²⁺ and/or are regulated by calcium-dependent processes, because of the link between the function of the 5-HT₃R in the hippocampus and cognitive processes. The injection of 5-HT₃R antagonists in freely moving rats facilitates the induction of long-term potentiation in the CA1 subfield of the hippocampus, and enhanced the retention of memory in hippocampal-dependent tasks (Staubli & Xu, 1995). These effects were shown to be due to a decrease in the firing activity of a subset of CA1 hippocampal interneurons, and a concomitant increase in the firing rate of hippocampal pyramidal cells (Reznic & Staubli, 1997).

At a variety of sites in the central and peripheral nervous systems, postsynaptic 5-HT₃Rs mediate fast synaptic transmission and presynaptic 5-HT₃Rs regulate neurotransmitter release (Jackson & Yakel, 1995; Roerig et al. 1997; Nayak et al. 1999). The function and regulation of the 5-HT₃Rs from these different locations can also vary. For example, the 5-HT₃Rs located on presynaptic rat striatal nerve terminals were found to be significantly more permeable to Ca²⁺ than classical postsynaptic types of 5-HT₃Rs (Rondé & Nichols, 1998), and to colocalize with the α4-nAChR subunit on a subset of these terminals (Nayak et al. 2000). The explanation for this functional diversity remains to be determined. The recent cloning of the 5-HT_3B subunit may help in understanding some of the functional diversity of 5-HT₃Rs. To date, however, no functional evidence that the 5-HT_3B subunit is participating in the formation of functional native 5-HT₃Rs exists. Furthermore, in the present experiments we could not detect the presence of the 5-HT_3B subunit (either functionally or molecularly) in rat hippocampal interneurons. It is highly likely that there are other subunits (and/or splice variants) that co-assemble with the 5-HT₃R, whether they are specific or not for the 5-HT₃R, which remain to be discovered.

Although MTSEA potentiated 5-HT-gated responses in Xenopus oocytes expressing both the 5-HT_3A and α4-nAChR subunits, and in half of the hippocampal interneurons studied, the extent of that potentiation was larger in the hippocampal neurons (i.e. 14 % vs. 52 %). There are several potential reasons for this apparent discrepancy. First, although we know that the α4-nAChR subunit does co-assemble with the 5-HT_3A subunit when both are expressed in oocytes or HEK-293 cells, the efficiency of this co-assembly is rather low (van Hooft et al. 1998; Kriegler et al. 1999a; Harkness & Millar, 2001). We previously estimated, using densitometric analysis of Western blots from HEK-293 cells co-expressing both 5-HT_3A and α4-nAChR subunits, that only 4-11 % of the 5-HT₃Rs contain the α4-nAChR subunit (van Hooft et al. 1998). Even though we injected a 50 times higher amount of α4-nAChR subunit RNA in the present experiments, it is still highly likely that the majority of the 5-HT₃Rs are homomeric assemblies of the 5-HT_3A subunit (Kriegler et al. 1999a). Nevertheless, this apparent low efficiency of co-assembly between the 5-HT_3A and α4-nAChR subunits does not necessarily have to hold true in native tissue. It is possible that cell-specific mechanisms exist that govern efficient co-assembly. Interestingly, although the α7- and α4β2-nAChRs form the major types of nAChRs in the brain (Jones & Yakel, 1997), these channels are poorly expressed in HEK-293 cells (Harkness & Millar, 2001). Another possible explanation of the lower extent of MTSEA-induced potentiation in oocytes is that the form of the 5-HT_3A subunit that we have used for these experiments is the mouse long form (Maricq et al. 1991); this could somehow affect the MTSEA-induced regulation. Finally, the mechanism whereby MTSEA potentiates α4-nAChR subunit-containing channels is unknown, and perhaps the effect of the MTSEA-induced regulation of ion channels is different between Xenopus oocytes and hippocampal interneurons.

In conclusion, we have shown that the primary 5-HT₃R subunit functionally expressed in rat hippocampal CA1 interneurons is the 5-HT_3A-short subunit; neither the 5-HT_3B subunit nor the 5-HT_3A-long subunit was detected. In addition, we found that the α4-nAChR subunit was co-expressed with the 5-HT_3A-short subunit, and have provided functional data consistent with the notion that this nAChR subunit co-assembles to form a functional channel with the 5-HT₃R. Such co-assembly provides a mechanism to explain functional diversity of the 5-HT₃R in the hippocampus and elsewhere in the brain, and perhaps for other neurotransmitter receptors.