Evidence for Expression of Heteromeric Serotonin 5-HT3 Receptors in Rodents

Authors

  • Michael C. Hanna,

  • Paul A. Davies,

  • Tim G. Hales,

  • Ewen F. Kirkness


  • Abbreviations used : GAPDH, glyceraldehyde-3-phosphate dehydrogenase ; HEK-293, human embryonic kidney 293 ; 5-HT, 5-hydroxytryptamine (serotonin) ; NGF, nerve growth factor ; SDS, sodium dodecyl sulfate ; SSC, saline-sodium citrate ; TC, tubocurarine ; TEN, Tris-HCl (10 mM), EDTA (1 mM), NaCl (100 mM), pH 7.5.

Address correspondence and reprint requests to Dr. E. F. Kirkness at The Institute for Genomic Research, 9712 Medical Center Drive, Rockville, MD 20850, U.S.A. E-mail : ekirknes@tigr.org

Abstract

Abstract : The gene and cDNAs that encode a novel subunit of rodent serotonin 5-HT3 receptors were isolated from mouse and rat tissues. Each of the new rodent subunits shares 40% amino acid identity with the rat 5-HT3A subunit and 73% identity with the human 5-HT3B subunit. Despite a relatively low level of structural conservation, sequence analysis and functional studies suggest that the new rodent subunits are orthologues of the human 5-HT3B subunit. In common with homologous human receptors, rat heteromeric 5-HT3 receptors displayed a substantially larger single-channel conductance than homomeric 5-HT3A receptors. In addition, the rat heteromeric receptors were less sensitive to antagonism by tubocurarine. However, in contrast to human heteromeric receptors, those of the rat displayed pronounced inward rectification of both the whole-cell and single-channel current amplitudes. Transcripts of the mouse 5-HT3A and 5-HT3B subunits are coexpressed in several cell lines that possess endogenous 5-HT3 receptors. In addition, treatment of rat PC12 cells with nerve growth factor induced expression of both subunit mRNAs, with a similar time course for accumulation of each transcript. The combination of functional data and expression patterns is consistent with the existence of heteromeric 5-HT3 receptors in rodent neurons.

The neurotransmitter serotonin [5-hydroxytryptamine (5-HT)] mediates rapid excitatory responses in peripheral and central neurons by activating ligand-gated ion channels (5-HT3 receptors). These receptors are expressed in a variety of peripheral ganglia, where they are thought to modulate responses to pain, and to control reflexes of the enteric and cardiovascular systems. In the central nervous system, 5-HT3 receptors have been implicated in the control of emesis, and antagonists of 5-HT3 receptors have found clinical use for suppression of the nausea that accompanies postoperative recovery and many cancer therapies (King et al., 1994).

Most families of ligand-gated ion channels are composed of multiple subunit types that assemble in alternative combinations to achieve functional diversity. For these receptor families, identification of the first subunit gene has usually led to the rapid discovery of multiple paralogues. However, the molecular characterization of 5-HT3 receptors has not followed this course. Although the first 5-HT3 receptor subunit gene was identified several years ago (5-HT3A ; Maricq et al., 1991), the first paralogue of this gene was described only recently (5-HT3B ; Davies et al., 1999).

Recombinant expression of 5-HT3A subunits alone yields functional 5-HT3 receptors (Maricq et al., 1991 ; Peters et al., 1997). However, heteromeric assemblies of the human 5-HT3A and 5-HT3B subunits more closely resemble native 5-HT3 receptors of rodent ganglia with respect to their single-channel conductance and permeability to Ca2+ ions (Davies et al., 1999). Consequently, it is likely that rodent ganglia normally express heteromeric 5-HT3 receptors. However, the 5-HT3 receptors of different species display several distinctive properties, particularly with respect to their pharmacological profiles (Peters et al., 1997). Previously, some of these differences have been explained by specific structural differences between the homomeric 5-HT3A receptors of different species (e.g., Hope et al., 1999). In this study, we have examined the functional properties of rat heteromeric 5-HT3 receptors and have defined important similarities and differences between heteromeric receptors of human and rat that can be attributed to the 5-HT3B subunit.

The detection of multiple 5-HT3 receptor subunits raises questions of how their expression is coordinated in vivo. It is known that neurotrophic factors can induce expression of functional 5-HT3 receptors in cultured nodose neurons (Rosenberg et al., 1997) or PC12 cells (Furukawa et al., 1992). It is therefore of interest to determine the effects of these factors on expression of the different 5-HT3 receptor subunits. Here we have examined the ability of nerve growth factor (NGF) to induce expression of the different subunit mRNAs in PC12 cells. The observed pattern of subunit expression suggests that PC12 cells will be a useful model system to dissect the mechanisms of regulated receptor expression.

MATERIALS AND METHODS

Isolation of the mouse and rat 5-HT3B subunit cDNAs

Exons of the rat and mouse 5-HT3B subunit genes were amplified by PCR from total genomic DNA using primers that were derived from the human 5-HT3B subunit cDNA sequence (Davies et al., 1999). The primers were nucleotides 435-474 (sense) and 564-587 (antisense) of GenBank accession no. AF080582. Amplification at 95°C for 45 s, 50°C for 60 s, and 72°C for 2 min was performed for 30 cycles using the XL-PCR system (Perkin-Elmer). Reaction products were purified from agarose gels and sequenced directly. Additional exons of the mouse 5-HT3B subunit gene were sequenced from larger fragments of genomic DNA that were cloned in phage λ (see below). Oligonucleotide primers were designed from the exon sequences to amplify the 5′ and 3′ flanking sequences from rat brain and mouse brain cDNA libraries, using the Marathon system (Clontech). Amplification at 95°C for 45 s, 60°C for 60 s, and 72°C for 2 min was performed for 35 cycles using the XL-PCR system. Reaction products were purified from agarose gels and sequenced directly. The rat and mouse 5-HT3B open reading frames were then amplified from the same cDNA libraries. For rat, the primers contained nucleotides 1-24 (sense) and 1,319-1,346 (antisense) of the rat 5-HT3B cDNA sequence (GenBank accession no. AF155044). For mouse, the primers contained nucleotides 1-25 (sense) and 1,349-1,373 (antisense) of the mouse 5-HT3B cDNA sequence (GenBank accession no. AF155045). Cloned products were sequenced over their entire length to ensure that no mutations had been introduced.

Isolation of the mouse 5-HT3B subunit gene

A library of mouse genomic DNA (strain 129SVJ), cloned in λ FIX II (Stratagene), was probed at moderate stringency [1 X saline-sodium citrate (SSC), 0.1% sodium dodecyl sulfate (SDS), 50°C] with 32P-labeled fragments of the human, rat, and mouse 5-HT3B cDNAs (nucleotides 55-1,393 of GenBank accession no. AF080582, nucleotides 1-233 of GenBank accession no. AF155044, and nucleotides 559-1,373 of GenBank accession no. AF155045). Ten hybridizing clones were obtained from ~1 × 106 plaques. Overlapping inserts were determined by restriction fragment mapping and Southern blotting. Exons and flanking introns were sequenced from templates of purified λ DNA using the Dye Terminator Cycle Sequencing system (PE Biosystems). The sequences of seven genomic fragments that contain all of the exons have been deposited with GenBank under accession nos. AF155046-AF155052.

Cell culture

The cell lines NG108-15 (American Type Culture Collection) and NCB-20 (Minna et al., 1975) were cultured in Dulbecco's modified Eagle's medium (Life Technologies), supplemented with fetal bovine serum (10%), pyridoxine-HCl (5 μM), hypoxanthine (0.1 mM), aminopterin (400 nM), thymidine (160 μM), penicillin (50 U/ml), and streptomycin (50 μg/ml). The cell lines N1E-115, human embryonic kidney 293 (HEK-293 ; American Type Culture Collection), and GT1-7 (Mellon et al., 1990) were cultured in Dulbecco's modified Eagle's medium, supplemented with calf serum (10%) and antibiotics. PC12 cells (American Type Culture Collection) were cultured on poly-L-lysine-coated plates, in RPMI medium, supplemented with horse serum (10%), fetal bovine serum (5%), and antibiotics. PC12 cells were induced to differentiate by addition of mouse 7S NGF (100 ng/ml ; Promega) to the culture medium.

Electrophysiology

Transient transfection of HEK-293 cells with subunit cDNAs was performed as described previously (Davies et al., 1999). The cDNAs encoding the rat 5-HT3A subunit (Mair et al., 1998) and rat 5-HT3B subunit (nucleotides 1-1,343) were each cloned in pCDM8 (Invitrogen). Cells were transfected with green fluorescent protein cDNA and 5-HT3 receptor cDNAs either alone or in combination. Whole-cell and outside-out patch configurations were used to record macroscopic and single-channel currents, respectively, from the membranes of fluorescent cells (Davies et al., 1999). The extracellular solution contained the following (in mM) : NaCl, 140 ; KCl, 4.7 ; MgCl2, 1.2 ; CaCl2, 2.5 ; glucose, 11 ; and HEPES/NaOH, 10 (pH 7.4). The electrode solution contained the following (in mM) : KCl, 140 ; MgCl2, 2.0 ; EGTA, 11 ; ATP (Mg2+ salt) 0.1 ; and HEPES/KOH, 10 (pH 7.4). Electrodes were coated with Sylgard (Dow Corning, Midland, MI, U.S.A.) to reduce noise due to electrode capacitance in recording from outside-out patches. Junction potentials were zeroed with an open electrode in the recording chamber before each experiment. The liquid junction potential was trivial (1.7 mV), and its inappropriate compensation was ignored. Cells were voltage-clamped with an electrode potential of -60 mV. To avoid desensitization, 5-HT was applied at 30-s intervals by brief pressure ejection (50-200 ms) with continuous bath perfusion (5 ml/min). Experiments were performed at room temperature (20-24°C). 5-HT-evoked currents were monitored using an Axopatch-200A amplifier, then low-pass filtered with a cutoff frequency of 2 kHz, and digitized using a DigiData 1200 interface (Axon Instruments) for acquisition onto the hard drive of a personal computer. Data were acquired and analyzed using pCLAMP7 software (Axon Instruments). Graphs were plotted and fitted using Slide Write Plus 4.0 software (Advanced Graphics Software Inc.). All data are expressed as the arithmetic means ± SEM. Student's t test was used to compare data sets.

Northern blot analysis

Total RNA was extracted from cultured cells with Trizol (Life Technologies), and poly(A)+ RNA was purified with Oligotex resin (Qiagen), using conditions specified by the manufacturers. Samples of ~4 μg of poly(A)+ RNA or 40 μg of total RNA were electrophoresed on a 1.0% formaldehyde agarose gel, transferred to nylon membranes, and hybridized with 32P-labeled fragments of the mouse 5-HT3A subunit cDNA (nucleotides 988-1,470 of GenBank accession no. X72395), mouse 5-HT3B subunit cDNA (nucleotides 968-1,373), rat 5-HT3A subunit cDNA (nucleotides 970-1,452 of GenBank accession no. D49395), rat 5-HT3B subunit cDNA (nucleotides 963-1,343), or human glyceraldehyde-3-phosphate dehydrogenase (GAPDH) cDNA (nucleotides 789-1,140) (Tokunaga et al., 1987). The blots were washed at 60°C in 0.1 × SSC, 0.1% SDS before exposure. Membranes were stripped of hybridized probe by boiling for 10 min in 0.5% SDS before rehybridization with a different probe.

Radioligand binding assays

PC12 cells were scraped in Tris-HCl (10 mM), EDTA (1 mM), NaCl (100 mM), pH 7.5 (TEN), centrifuged (1,500 g, 5 min, 4°C), resuspended in TEN, and hand-homogenized. Homogenates were centrifuged (50,000 g, 30 min, 4°C), and the membrane pellet was resuspended in TEN at a concentration of 10 mg of protein/ml. Binding activity was measured by incubating membrane protein (200 μg) with [3H]GR65630 (1 nM ; 63 Ci/mmol ; New England Nuclear) for 60 min at 25°C in 500 μl of Tris-HCl (50 mM), EDTA (1 mM), pH 7.5. Nonspecific binding was defined in the presence of 3-tropanyl-3,5-dichlorobenzoate (MDL-72222 ; 1 μM ; Research Biochemicals International). Membrane-bound ligand was recovered by collection on GF/B filters under vacuum.

RESULTS

Cloning of mouse and rat subunit cDNAs

The sequence of the human 5-HT3B subunit cDNA (Davies et al., 1999) was used to design oligonucleotides for amplification of homologous exons from mouse and rat genomic DNA (see Materials and Methods). Exons of a homologous mouse gene were also located on hybridizing clones of genomic DNA (see below). The sequences of these exons were used to design primers for amplification of flanking cDNA fragments from mouse and rat brain cDNA libraries. Oligonucleotides that flanked the complete open reading frames of the mouse and rat cDNAs were then used to amplify and clone contiguous cDNAs that encode the complete mouse and rat subunits (Fig. 1A). The amino acid sequences of the mouse and rat subunits are 95% identical to each other and 73% identical to the human 5-HT3B subunit. The level of sequence similarity between the rodent and human subunits is relatively low when compared with subunits of related ligand-gated ion channels (generally 80-98% identity). However, on the basis of both sequence analysis and functional studies, it was concluded that the cloned rodent subunits are true orthologues of the human 5-HT3B subunit (see below and Discussion). Most of the variation between the rodent and human subunits is concentrated at the N-terminus and in the putative intracellular domain, between M3 and M4 (Fig. 1A). In other regions, including the unusual M2 domain (Davies et al., 1999), there is a high degree of sequence conservation between the three species.

Figure 1.

Amino-acid sequences and gene structure for rodent 5-HT3B subunits. A : Alignment of amino-acid sequences for the human, mouse, and rat 5-HT3B subunits. Conserved residues are boxed, and the four putative transmembrane domains (M1-M4) are highlighted by lines over the corresponding peptide sequences. Segments of the mouse subunit that are encoded by distinct exons are indicated by the location of exon termini (▾). B : Gene structure for the mouse 5-HT3B subunit. A composite map of the 5-HT3B subunit gene is shown above the three cloned gene fragments from which it was derived. The gene is illustrated with the locations of all protein-coding exons (filled boxes ; 1-9) and restriction endonuclease sites for KpnI (K) and EcoRI (E).

FIG. 1.

For both rodent cDNAs, the first ATG codon of the open reading frame is located downstream of an in-frame stop codon. This predicted site for initiation of translation corresponds to the second in-frame ATG codon (Met6) of the human 5-HT3B cDNA (Davies et al., 1999). A comparison of the flanking sequences for the first two in-frame ATG codons of the human cDNA shows that the second ATG codon more closely resembles the consensus sequence for initiation of translation (Kozak, 1991). It is therefore likely that this second ATG codon represents the true initiation site for translation of the human subunit. The human sequence of Fig. 1A has been numbered from the second in-frame methionine residue of the translated cDNA and is therefore five amino acid residues shorter than reported previously (Davies et al., 1999 ; Dubin et al., 1999).

Cloning of the mouse 5-HT3B subunit gene

Fragments of human, rat, and mouse 5-HT3B subunit cDNAs were used to screen a library of mouse genomic DNA. From 10 hybridizing plaques, the complete mouse 5-HT3B subunit gene was located within the inserts of three overlapping clones (Fig. 1B). The mouse 5-HT3B subunit is encoded by nine exons that span 28 kb of genomic DNA. The gene structure is identical to that of the mouse 5-HT3A subunit (Uetz et al., 1994), with all exon/intron boundaries located in equivalent positions of homologous codons. This conserved structure supports the view that the two subunit genes are derived from a common ancestral gene that underwent a local duplication event (Davies et al., 1999).

Functional properties of recombinant rat 5-HT3 receptors

The functional properties of recombinant rat 5-HT3 receptors were determined in voltage-clamped HEK-293 cells. Application of 5-HT (100 μM) did not induce currents in cells transfected with the rat 5-HT3B cDNA alone. However, transfection of the 5-HT3A cDNA in the absence or presence of the 5-HT3B cDNA resulted in expression of 5-HT3 receptors with distinctive sensitivities to agonist. Application of 5-HT (0.3-100 μM) caused a concentration-dependent activation of currents recorded from cells transfected with the 5-HT3A cDNA alone (EC50 = 1.4 ± 0.1, Hill slope = 1.6 ± 0.2) or cDNAs encoding both 5-HT3A and 5-HT3B subunits (EC50 = 2.4 ± 0.5, Hill slope 1.2 ± 0.3). The 1.7-fold shift of the EC50 value for 5-HT that is induced by expression of the rat 5-HT3B subunit is similar to that found with human heteromeric receptors (2.1-fold ; Davies et al., 1999). The ability of tubocurarine (TC) to inhibit these responses was of interest owing to the differential potency of this antagonist at rodent and human 5-HT3A receptors (Mair et al., 1998) and its differential potency at human homomeric and heteromeric 5-HT3 receptors (Davies et al., 1999). Here, for both rat receptor preparations, the bath application of TC (3 nM to 1 μM) caused a concentration-dependent inhibition of the 5-HT-evoked current amplitude that reversed on washout of the antagonist (Fig. 2A). In common with human 5-HT3 receptors (Davies et al., 1999), cells expressing the homomeric and heteromeric subunit combinations displayed differential sensitivity to TC, with IC50 values of 14.5 ± 1.8 and 27.6 ± 1.9 nM, respectively (significantly different at p < 0.05).

Figure 2.

TC sensitivity and current-voltage relationship of rodent homomeric and heteromeric 5-HT3 receptors. A : The concentration dependence of TC-evoked inhibition of 5-HT (10 μM)-activated whole-cell currents recorded from HEK-293 cells expressing either 5-HT3A (•) or both 5-HT3A and 5-HT3B subunit cDNAs (○). The IC50 values for inhibition of homomeric and heteromeric receptors were significantly different : 14.5 ± 1.8 and 27.6 ± 1.9 nM, respectively (p < 0.05). Data points are mean values of at least four experiments. B : There was no difference between the current-voltage relationships of whole-cell currents recorded from HEK-293 cells expressing homomeric and heteromeric receptors. Both combinations mediated inwardly rectifying currents that reversed in sign at 0 mV (n = 3). Vertical bars represent ±SEM.

FIG. 2.

Human homomeric and heteromeric 5-HT3 receptors display distinctive steady-state current-voltage relationships. Currents that are mediated by human homomeric 5-HT3A receptors exhibit pronounced inward rectification, whereas those of heteromeric receptors vary linearly with voltage (Davies et al., 1999 ; Dubin et al., 1999). However, this distinguishing feature was not displayed by recombinant rat 5-HT3 receptors. Currents recorded from cells that expressed the rat 5-HT3A subunit either alone or in combination with the rat 5-HT3B subunit could not be differentiated by their relationship to voltage (Fig. 2B). The current-voltage relationship for both receptor preparations exhibited a similar equilibrium potential (~0 mV) and pronounced inward rectification.

The most striking difference between human homomeric and heteromeric 5-HT3 receptors relates to their single-channel conductances, which vary by ~40-fold (Davies et al., 1999). It was therefore of interest to compare the single-channel conductances of the recombinant rat 5-HT3 receptors. 5-HT (10 μM) was applied to outside-out membrane patches after excision from cells expressing the rat 5-HT3A subunit either alone or in combination with the rat 5-HT3B subunit. For patches of homomeric 5-HT3A receptors clamped at - 140 mV, the activated currents rose and decayed relatively slowly and smoothly (Fig. 3A). These currents resembled miniature whole-cell currents, and single-channel events could not be resolved. In contrast, at the same holding potential and amplification, single-channel events were clearly resolved in patches from cells expressing both 5-HT3A and 5-HT3B subunits. The majority of openings had similar amplitudes as illustrated by the recording at a holding potential of -200 mV in Fig. 3B. Occasional openings occurred with a lower conductance. One such event can be seen in the current trace recorded at - 160 mV (Fig. 3A). These single-channel currents exhibited pronounced inward rectification (Fig. 3C). The cord conductance declined from 14 pS at a holding potential of -160 mV to 7 pS at -80 mV (Fig. 3D). It was not possible to measure the amplitude of single-channel events at potentials less hyperpolarized than -80 mV due to the reduction in the signal-to-noise ratio caused by inward rectification. By contrast, due to the more linear relationship between single-channel cord conductance and voltage in recordings from human heteromeric receptors, single-channel current amplitudes could be evaluated even at -40 mV (Davies et al., 1999).

Figure 3.

Currents recorded from outside-out patches containing homomeric and heteromeric rat 5-HT3 receptors. A : No discrete single-channel events were detectable in recordings of 5-HT (10 μM)-activated currents from outside-out patches containing 5-HT3A receptors. In contrast, at a similar amplification and holding potential, single-channel events could be resolved clearly in recordings from a patch containing several heteromeric 5-HT3 receptors held at the same potential (-140 mV). Increasing the driving force, by holding the patch at -160 mV, increased the amplitude of single-channel currents. In this recording, a rare subconductance event is denoted by the asterisk. B : Currents through heteromeric channels were recorded ~300 ms after application of 5-HT (10 μM), from a patch clamped at -200 mV. C : A three-dimensional plot of current amplitude and holding potential versus the number of single-channel events illustrates the pronounced inward rectification of the single-channel current amplitude. D : Cord conductances were derived from gaussian distributions fitted to the amplitude histograms in C. The data points, fitted with a polynomial, demonstrate the voltage dependence of the single-channel conductance.

FIG. 3.

Expression of the 5-HT3B subunit mRNA in rodent cell lines

Rodent cell lines that are known to express functional 5-HT3 receptors (N1E-115, NCB-20, NG108-15, and NGF-treated PC12 ; King et al., 1994) each contain an mRNA species of ~2.3 kb that hybridized at high stringency with a fragment of the 5-HT3B subunit cDNA (Fig. 4, top panels). The northern analysis did not detect this transcript in the mouse neural cell line, GT1-7, or in untreated PC12 cells (Fig. 4). Two other rodent cell lines of neural origin, Neuro-2A (Olmsted et al., 1970) and C6 (Benda et al., 1968), also lacked detectable levels of 5-HT3B subunit mRNA (data not shown). Sequencing of the 5-HT3B subunit cDNA from N1E-115 cells revealed an identical sequence to that obtained from mouse brain.

Figure 4.

Expression of 5-HT3 subunit mRNAs in rodent cell lines. The top panels show hybridization of 32P-labeled fragments of the 5-HT3B subunit cDNA to poly(A)+ RNA from NG108-15 cells (lane 1), NCB-20 cells (lane 2), N1E-115 cells (lane 3), GT1-7 cells (lane 4), untreated PC12 cells (lane 5), and PC12 cells that were treated for 9 days with NGF (lane 6). The same blots were stripped and reprobed with 32P-labeled fragments of the 5-HT3A subunit cDNA (middle panels) and GAPDH cDNA (bottom panels).

FIG. 4.

Cell lines that express functional 5-HT3 receptors are known to contain transcripts of the 5-HT3A subunit (Maricq et al., 1991 ; Isenberg et al., 1993), and this was confirmed here (Fig. 4, middle panels). However, the only description of 5-HT3A subunit mRNA in PC12 cells has indicated that untreated cells also contain a significant level of the transcript (Isenberg et al., 1993). In contrast to these data, we found no evidence for the expression of 5-HT3A subunit mRNA in untreated PC12 cells (Figs. 4 and 5A). Transcripts of each 5-HT3 receptor subunit were detectable by northern analysis only after 3 days of treatment with NGF. The study of Isenberg et al. (1993) quantified the levels of 5-HT3A subunit mRNA in PC12 cells by PCR amplification after reverse transcription of cellular RNA with random primers. If untreated PC12 cells contain transcripts of the 5-HT3A subunit that are not polyadenylated, it is possible that such transcripts would be detected by this approach, but not by northern analysis of poly(A)+ mRNA. To test this possibility, northern analysis was also performed on total cellular RNA from PC12 cells. In common with the results obtained with poly(A)+ mRNA, hybridizing subunit mRNAs were detectable only after at least 3 days of treatment with NGF (Fig. 5A, lanes 6-10).

Figure 5.

Time dependence for expression of 5-HT3 receptors in NGF-treated PC12 cells. A : Northern blot analysis of poly(A)+ RNA (lanes 1-5) or total RNA (lanes 6-10) from PC12 cells that were untreated (lanes 1 and 6) or were treated with NGF for 1 day (lanes 2 and 7), 3 days (lanes 3 and 8), 6 days (lanes 4 and 9), and 9 days (lanes 5 and 10). The top panels show hybridization of 32P-labeled fragments of the rat 5-HT3A subunit cDNA. The same blots were stripped and reprobed with 32P-labeled fragments of the rat 5-HT3B subunit cDNA (middle panels) and GAPDH cDNA (bottom panels). B : Binding of [3H]GR65630 to membranes of PC12 cells that were untreated or were treated with NGF for up to 9 days. Data are the means ± SD of three independent experiments.

FIG. 5.

Expression of receptor protein in PC12 cells was monitored by the binding of [3H]GR65630. Consistent with the results of the northern blot analysis, radioligand binding was detectable only after 3 days of treatment with NGF (Fig. 5B). Furthermore, in whole-cell recordings, at a holding potential of -60 mV, 5-HT (10 μM) failed to activate discernible currents in untreated PC12 cells, but induced robust currents in NGF-differentiated PC12 cells (n = 7 ; data not shown).

DISCUSSION

Here we report the isolation of genomic DNA and cDNAs that encode rodent homologues of the human 5-HT3B subunit. These subunits each share only 73% amino acid sequence identity with their human homologue and are therefore markedly less well conserved than the rodent and human orthologues of the 5-HT3A subunit (85% identity). This relatively low level of sequence conservation raises the question of whether the genes described in this study are true orthologues of the human 5-HT3B subunit or represent a distinct subunit type. Here their classification as 5-HT3B orthologues was based on several criteria. First, screening of a mouse genomic library at low stringency with the human 5-HT3B subunit cDNA failed to reveal any cross-hybridizing genes that were more similar to the human subunit than the gene described here. Second, the presumptive promoter elements in the proximal 5′-regions of the mouse and human 5-HT3B subunit genes display a very high level of sequence conservation (80% nucleotide identity over 450 bp ; data not shown). Finally, the rat and human 5-HT3B subunits confer similar functional properties to recombinant 5-HT3 receptors.

Relative to the 5-HT3A subunit, the overall level of sequence conservation for the 5-HT3B subunits of different species indicates that its biological functions can be maintained by fewer conserved amino acid residues. Notably, the conserved residues include those of the M2 region that distinguish the 5-HT3B subunit from all other related receptor subunits (Davies et al., 1999). Before the identification of 5-HT3B subunits, it was suggested that the ligand-binding properties of native 5-HT3 receptors are determined solely by the 5-HT3A subunit, whereas additional subunits affect only the biophysical response of the channel (Fletcher and Barnes, 1998). The functional properties of homomeric and heteromeric 5-HT3 receptors in recombinant systems (Davies et al., 1999 ; Figs. 2 and 3) are generally consistent with this proposal. It may also explain why, outside of the channel-forming domains, the 5-HT3B subunit has evolved under less constraint than the 5-HT3A subunit.

In common with recombinant human 5-HT3 receptors, coexpression of rat 5-HT3A and 5-HT3B subunits yielded receptors that can be distinguished from homomeric 5-HT3A receptors by their greatly enhanced single-channel conductance and their reduced sensitivity to TC. The single-channel conductance of heteromeric receptors was resolved readily in isolated membrane patches and was of similar magnitude to that displayed by 5-HT3 receptors of rodent ganglia (Yang et al., 1992 ; Hussy et al., 1994). The effects of TC at the different rat 5-HT3 receptors are of particular interest owing to the unusual potency of this antagonist at rodent 5-HT3 receptors. Homomeric 5-HT3A receptors of rat and human differ in their sensitivity to TC by ~100-fold (Mair et al., 1998). Inclusion of a 5-HT3B subunit in the recombinant receptors caused only a modest reduction of TC potency in both human (approximately fivefold ; Davies et al., 1999) and rat (approximately twofold ; Fig. 2). Heteromeric receptors from the different species therefore retained their large differential sensitivity to TC.

A notable difference between the heteromeric 5-HT3 receptors of human and rat relates to the voltage dependence of channel function. The inward rectification that is displayed by human homomeric 5-HT3A receptors was not observed in similar recordings from cells expressing human heteromeric receptors (Davies et al., 1999 ; Dubin et al., 1999). In contrast, both homomeric and heteromeric rat receptors exhibited inward rectification. The inward rectification of heteromeric channels can be explained, at least in part, by rectification of the single-channel cord conductance (Fig. 3C). Notably, inward rectification of 5-HT-activated single-channel currents has been observed previously in recordings from rat neurons (Yang et al., 1992).

Transcripts of the 5-HT3B subunit gene were detectable in three cell lines (N1E-115, NCB-20, and NG108-15) that have been used extensively for characterization of functional 5-HT3 receptors. The subunit mRNA is not expressed ubiquitously and was not detected in several cell lines of neural origin that lack 5-HT3 receptors. Although the mouse 5-HT3B subunit has not yet been examined in recombinant systems, the properties of recombinant 5-HT3 receptors from human (Davies et al., 1999) and rat (Fig. 3) predict that cell lines that express both 5-HT3 receptor subunits should be capable of assembling 5-HT3 receptors that display a large single-channel conductance. Such receptors have been reported in N1E-115 cells (van Hooft and Vijverberg, 1995) and NG108-15 cells (Shao et al., 1991). However, several other studies with these cell lines have observed single-channel conductance values only in the sub-picosiemens range (e.g., Lambert et al., 1989 ; Hussy et al., 1994). The reason for these discrepancies is unclear at present. It is possible that the assembly and transport of heteromeric 5-HT3 receptors require auxiliary factors that are absent from some lines of the mouse neuroblastoma cells. Evidence that both low- and high-conductance 5-HT3 receptors can be expressed within a single neuron (Hussy et al., 1994) suggests that some neurons may assemble both homomeric and heteromeric receptors. If so, it is likely that specialized auxiliary factors are required to segregate the assembly and transport of the different receptor types. However, testing the idea of differential trafficking for the different 5-HT3 receptor subunits in neurons, and in cell lines, will require suitable antisera with which to track the subunit proteins.

A correlation between the assembly of functional 5-HT3 receptors and expression of both subunit mRNAs was observed in PC12 cells. Previously, the only description of 5-HT3A subunit mRNA in PC12 cells had indicated that untreated cells also contain a significant level of the transcript (Isenberg et al., 1993). In that study, a basal level of 5-HT3A subunit mRNA was increased by up to fourfold following treatment with NGF for 8 days. It was argued that this increase could account for the concurrent appearance of functional 5-HT3 receptors. However, in contrast to these data, we found no evidence for expression of 5-HT3A subunit mRNA in untreated PC12 cells. Transcripts of each 5-HT3 receptor subunit were detectable by northern analysis only after 3 days of treatment with NGF. The absence of subunit transcripts in untreated PC12 cells is more consistent with the lack of detectable 5-HT3 receptors in these cells (Gordon and Rowland, 1990 ; Furukawa et al., 1992 ; Isenberg et al., 1993). In this study, receptor expression was monitored by the binding of [3H]GR65630 and was also detectable only after 3 days of treatment with NGF. The combined data demonstrate a clear temporal correlation between expression of both subunit mRNAs and the assembly of 5-HT3 receptors at the cell membrane of PC12 cells.

The 5-HT3 receptors of PC12 cells are not well characterized, relative to those of mouse neuroblastoma cells. However, PC12 cells are likely to represent a valuable model system in which to examine several properties of 5-HT3 receptors. They respond to NGF by differentiating into cells that resemble sympathetic neurons (Greene and Tischler, 1976) and can thereby provide a relevant environment for the study of peripheral 5-HT3 receptors. In addition, it is likely that neurotrophic factors, including NGF, are necessary for induction and maintenance of 5-HT3 receptor expression in vivo (Rosenberg et al., 1997). The effects of NGF on 5-HT3 receptor expression in PC12 cells may therefore reflect a relevant physiological process. An attractive feature of PC12 cells is the absence of detectable receptor expression in untreated cells, which permits unambiguous identification of factors that cause induction of receptor expression. This contrasts with most other known ion channels of PC12 cells, in which NGF causes only a change in the level of constitutive gene expression (e.g., Rogers et al., 1992 ; Sharma et al., 1993 ; Casado et al., 1996). This property will be of particular value for dissection of the intracellular signaling pathways that control expression of native 5-HT3 receptors.

The functional characterization of novel rodent 5-HT3 receptor subunits has demonstrated that several properties of 5-HT3 receptors in rodent neurons (Yang et al., 1992 ; Hussy et al., 1994) are best explained by the existence of heteromeric receptors. Such receptors are also predicted by our preliminary studies that have detected expression of rat 5-HT3B subunit mRNA in tissue regions (e.g., amygdala and peripheral ganglia) where transcripts of the 5-HT3A subunit are coexpressed (unpublished observations). Although rat and human 5-HT3B subunits share only 73% amino acid sequence identity, they each confer a greatly increased single-channel conductance and a reduced sensitivity to antagonism by TC. However, rat and human heteromeric receptors also exhibit a marked difference in their voltage dependence. The sequences and expression patterns described here can now be used to define the molecular determinants of these properties and to explore the possibility of heterogeneous 5-HT3 receptors in rodent tissues.

Ancillary