P2X receptor signaling inhibits BDNF-mediated spiral ganglion neuron development in the neonatal rat cochlea

In rodent models, such as the rat, where detection of airborne sound commences at around postnatal day 11 (P11)(Geal-Dor et al., 1993), it has been established that the afferent innervation of the cochlea arises from a programmed pattern of neurite outgrowth to the sensory hair cells, followed by selective pruning in the week prior to the onset of hearing(Simmons et al., 1991). The molecular mechanism for rationalization of afferent neurite extension and branching that leads to the mature cochlear innervation pattern has not been established, but occurs in the face of ongoing trophic support from the hair cell-derived neurotrophins.

In the mature mammalian cochlea, 90-95% of the primary afferent fibers innervate individual inner hair cells (IHCs). The remaining 5-10% of the cochlear afferents have neurites that branch extensively to synapse with multiple outer hair cells (OHCs). These are the type I and type II spiral ganglion neurons (SGN), respectively(Berglund and Ryugo, 1987). The IHC-type I SGN pathway represents the principal channel for sound transduction and auditory neurotransmission, with each IHC supporting exclusive synapses with many type I SGN. The function of the OHC-type II SGN pathway has not been determined but is likely to provide sensory feedback from the cochlear amplifier, which is an active tuning process that supports the sensitivity and frequency selectivity of the hearing organ (e.g. Jagger and Housley, 2003). Thus, the more numerous type I SGN initially innervate both IHC and OHC, and subsequent pruning of these connections results in withdrawal of redundant neurites from the OHC and refinement of the innervation at the IHC. Type II SGN initially innervate both IHC and OHC, with subsequent loss of the IHC connections (see Fig. 1) (for reviews, see Pujol et al.,1998; Rubel and Fritzsch,2002). Neurotrophins, particularly brain-derived neurotrophic factor (BDNF) and NT3, are secreted by the hair cells and act as survival factors for SGN. They may also promote neurite extension and branching during synaptogenesis in the cochlea (Mou et al.,1997; Pirvola and Ylikoski,2003). Expression of the neurotrophins BDNF and NT3 by the hair cells and their respective receptors, TRKB and TRKC, by the SGN is established as neurite outgrowth from the spiral ganglion commences (at around E18.5) and is sustained through the neurite pruning that establishes the mature cochlear afferent innervation pattern (Pirvola and Ylikoski, 2003).

Candidates for signal transduction pathways that contribute to regulation of axon growth and synaptic determination include traditional axon guidance molecules, and also several neurotransmitters such as glutamate, acetylcholine and adenosine 5′-triphosphate (ATP)(Huang et al., 2006). ATP mediates cell signaling through activation of two classes of purinergic receptor, the metabotropic P2Y receptor and the ionotropic P2X receptor. The P2X receptor family (P2X₁-P2X₇) has wide distribution among neural systems, including the inner ear(Khakh and North, 2006). ATP-gated ion channel diversity arises from heteromeric assembly and alternative splicing of the P2X receptor subunits. P2X₃ receptors have been shown to inhibit motor axon outgrowth in neural tube explants(Cheung et al., 2005). We have now established that P2X₃ receptor expression is prominent in rat and mouse SGN and has a spatiotemporal expression pattern that matches the early postnatal period of neurite reorganization(Huang et al., 2005; Huang et al., 2006). Trafficking of the P2X₃ receptor protein matched the timing of cochlear afferent synaptic restructuring prior to the onset of hearing. In the first few postnatal days in the rat, P2X₃ immunolabeling in the neurite terminals was confined to the inner spiral plexus-IHC region, then extended to include tunnel-crossing afferents innervating the OHC. P2X₃ expression was then downregulated under the IHC by P8, and lost from the outer spiral bundle fibers innervating the OHC by P14(Huang et al., 2005). This contrasts with the sustained P2X₂ receptor expression in both the inner radial fibers and the outer spiral fibers in the developing rat cochlea(Järlebark et al., 2000). The other five P2X receptor subunits in the rat SGN exhibit variable expression (Xiang et al.,1999; Nikolic et al.,2001; Nikolic et al.,2003), with P2X₁ downregulating from P2(Nikolic et al., 2001). To date, no other candidate neurohumoral signal transduction pathways such as the glutamate receptors or neurotrophin receptors show the coincident expression profile of the P2X₃ receptor. P2X receptors containing the P2X₃ subunit therefore represent a candidate molecular signaling pathway, which may complement neurotrophic support and reshape the afferent innervation of the cochlear hair cells.

Spiral ganglion neurons of the neonatal rat cochlea express ATP-gated ion channels with a unique P2X receptor phenotype. Features of this P2X receptor include unusual sensitivity to ATP and broad agonist action, which shows some semblance of both P2X₂ and P2X₃ phenotype but cannot be readily reconciled with published data for recombinant P2X_2/3receptor heteromers (Salih et al.,2002). P2X₂, P2X₃ and P2X_2/3receptors contribute to transduction and neurotransmission in a range of sensory modalities (for reviews, see Khakh and North, 2006; North,2002), with alternative splicing potentially broadening receptor diversity (Brändle et al.,1997; Salih et al.,1998).

The present study undertook a molecular characterization of the P2X receptors expressed by the rat neonatal SGN during the crucial period for neural reorganization immediately prior to the onset of hearing. Using quantitative single-cell RT-PCR, we found that the P2X₃ subunit was the dominant transcript, with approximately 50% greater abundance than the P2X₂ subunit; compatible with a P2X₃-P2X₃-P2X₂ trimer assembly(Jiang et al., 2003). The other five P2X receptor transcripts were less abundant. Alternative splicing of the P2X₂ receptor was apparent, with dominance of a P2X_2-3 isoform(Salih et al., 1998). The pharmacological properties of the recombinant P2X_2-3/3 heteromer, expressed in Xenopus oocytes, matched the native SGN P2X receptor phenotype. This pharmacological profile was exploited in a neonatal spiral ganglion explant model to demonstrate that ATP, acting through this P2X receptor, counters the neurite extension and branching elicited by the neurotrophin BDNF. These findings establish a mechanism contributing to the pruning and withdrawal of neurites contacting the sensory hair cells, which is required for the reorganization of cochlear synaptic innervation as hearing is established.

All procedures were performed with the approval of the University of Auckland Animal Ethics Committee or the Animal Subjects Committee of the San Diego VA Medical Center.

Total RNA was extracted from spiral ganglia dissected from cochleae of Wistar rats at postnatal ages P0-P4 inclusive and P14, using Trizol(Invitrogen). Reverse transcription was performed on 0.5 μg total RNA using random hexamers and Superscript II (Invitrogen) in a 20 μl reaction mix. A non-quantitative PCR amplification for all the P2X receptors was performed on cDNA samples (4 μl) in a reaction volume of 50 μl containing 1×PCR buffer, 2 mM MgCl₂, 200 μM dNTPs, 200 nM each of forward and reverse primers and 1 U AmpliTaq Gold DNA polymerase (Perkin-Elmer). PCR amplification included denaturation at 96°C for 5 minutes, then 48 cycles of 94°C for 45 seconds, 58°C for 45 seconds and 72°C for 90 seconds, followed by 72°C for 10 minutes. Samples (10 μl) of PCR product were analysed by agarose gel electrophoresis. Each experiment consisted of the P2X receptor target in duplicate. Controls included omission of the reverse transcriptase reaction or no cDNA template. The procedure was repeated using cDNA derived from two independent rat spiral ganglia (a summary of PCR primers is given in Tables 1 and 2).

Cochlear slices were prepared from neonatal rats (P2-P4) as described by Jagger et al. (Jagger et al.,2000). SGN were aspirated into glass micropipettes (GC120TF-10;Clark Electromedical Instruments, UK) containing 2 μl of sterile 0.1 M PBS. In some instances cells were aspirated into recording pipettes following voltage-clamp analysis of P2X receptor currents (see `patch clamp of SGN'section). The contents of the electrode were immediately expelled into a chilled, thin-walled silicon-coated 0.5 ml microfuge tube (Invitrogen)containing 9 μl of RT reaction mixture (1× RT buffer, 150 ng random hexamers, 1 mM dNTPs, 20 U RNAseOut; Invitrogen). Cells were stored at-80°C for 1 hour and then thawed on ice to promote lysis. Cells were incubated at 65°C for 10 minutes, then cooled to 25°C before the addition of 0.5 μl Superscript II reverse transcriptase. Reverse transcription proceeded for 10 minutes at 25°C, then 1 hour at 42°C,followed by inactivation at 70°C for 10 minutes. cDNAs not used immediately were stored at -80°C. Controls included RT-PCR of bath solution collected adjacent to SGN, no template controls and omission of reverse transcription for SGN mRNA samples.

Real-time quantification of mRNAs for P2X receptor and the control genes glyceraldehyde-3-phosphate dehydrogenase (Gapdh) and neuron-specific enolase [(Nse), also known as Eno2 - Mouse Genome Informatics] was performed using gene-specific primers and TaqMan fluorogenic MGB probes. The oligonucleotide primers and TaqMan MGB probes (Tables 1 and 2) were designed using Primer Express v1.5 (PE Applied Biosystems). The real-time PCR reaction was performed using an ABI 7700 Sequence Detection System (PE Applied Biosystems). The reaction mixture (25 μl) consisted of a 2 μl sample of neuron cDNA (as a 1:1 dilution of the first-strand cDNA synthesis reaction, see above), 750 nM of each primer, 200 nM of TaqMan MGB probe and 1× TaqMan Universal PCR MasterMix (PE Applied Biosystems). PCR amplification was performed for 2 minutes at 50°C, heated to 95°C for 10 minutes and then followed by 50 cycles of 15 seconds at 95°C and 1 minute at 60°C. Standard curves for each gene target were generated by amplification using conventional PCR to produce larger amplicons flanking the real-time PCR targets. These PCR products were purified by Qiagen gel extraction (Qiagen) from 1.5% agarose gels. The DNA concentration (ng/μl) was determined by UV spectrometry and the starting copy number calculated, providing the basis for serial dilution down to single-copy equivalence. Standard curves were derived from plotting cycle threshold (C_T) against copy number (using the 10,000 to single-copy standard dilution series) in triplicate. Each experiment contained a series of negative and positive controls, including no template controls,bath solution, negative tissues and omission of reverse transcription. Transcript numbers were expressed as the mean±s.e.m. Statistical significance among different P2X receptor transcript numbers was assessed by one-way analysis of variance (ANOVA) and post-hoc Student's paired t-tests.

The single-cell real-time PCR method employed here was designed to allow quantification of nine different gene transcripts from individual SGN at low copy number. This approach required division of the cDNA derived from each neuron to provide a sufficient template for multiple PCRs. Using this approach, the relative abundance and not the absolute copy number of each transcript was sought, such that our methodology establishes the proportional representation of specific transcripts relative to each other, based on real-time PCR utilizing TaqMan chemistry(Bustin, 2002).

In control experiments we initially confirmed the specificity of the primer/probe sets by performing PCRs using specific P2X receptor subunit cDNA templates with all primer/probe combinations. The lack of amplification of off-target templates up to 50 cycles confirmed the specificity of the PCRs(Fig. 2A,B). Standard curves using dilutions of the cDNA templates confirmed reproducibility and sensitivity of the PCR protocol. A standard curve of copy number against C_T value was generated from these data and the linear regression equation y=mx+c was derived for quantification of single-cell P2X receptor gene copies. This analysis showed that the standard curves from the dilution series were linear to <10 copies(Fig. 2C). Analysis of each dilution series enabled determination of the detection limit, slope and PCR efficiency of each reaction (Fig. 2D) and therefore the reproducibility and reliability of using interpolation from the standard curves to quantify multiple gene transcripts amplified from the same neuron. For each target, the slope of the standard curve approximated -3.32, the theoretical value representing a PCR reaction where the number of template molecules double with each cycle and hence are 100% efficient (Ginzinger,2002; Pfaffl et al.,2002).

A first round of PCR amplification was performed in 50 μl using 5 μl of cDNA template, with 1× PCR buffer, 2 mM MgCl₂, 200 mM dNTPs, 10 nM primers and 1 U AmpliTaq Gold polymerase (PE Applied Biosystems)for 30 cycles of 30 seconds at 94°C, 45 seconds at 60°C and 90 seconds at 72°C, followed by 7 minutes at 72°C. A second round of PCR was performed using 4 μl of first-round product as a template (final volume 25μl). In this second PCR, the P2X₂ splice variants were amplified using a second set of internal primers (100 nM each) and by performing 46 PCR cycles as described above. The internal PCR primers designed for P2X₂ (Accession No. U14414) semi-nested single-cell PCR are included in Table 1. Neurons were processed in batches of six, and positive and negative controls were included. Amplicons were verified by agarose gel electrophoresis.

Cochleae dissected from neonatal P4 rat pups were fixed in 4% formaldehyde and 0.5% gluteraldehyde in 0.1 M phosphate buffer (pH 7.4) for the P2X₂ antibody and 4% paraformaldehyde in 0.1 M phosphate buffer (pH 7.4) for the P2X₃ and P2X₄ antibodies. Tissues were cryoprotected and sectioned as previously described(Huang et al., 2005; Järlebark et al., 2000). After a series of washes in 0.1 M PBS, sections were incubated for 2 hours at room temperature in a blocking-permeabilization solution of 10% normal goat serum (Vector Laboratories) and 1% Triton X-100 in 0.1 M PBS. Primary antibody was applied in 10% normal goat serum and 0.1% Triton X-100 at dilutions of 1:500 for P2X₂, 1:30,000 for P2X₃ and 1:200 for P2X₄ and incubated overnight at 4°C. Primary antibody was removed by 0.1 M PBS washes. Sections were incubated with secondary antibody(Alexa Fluor-594) at 1:500 for 2 hours at room temperature and 2 hours at 4°C. Secondary antibody was removed by 0.1 M PBS washes as described but with an additional incubation in PBS overnight at 4°C. Sections were mounted in Citifluor (Agar Scientific, UK) on microscope slides (ProbeOn Plus;Fisher Scientific, Pittsburgh, PA). Immunofluoresence images were obtained by confocal microscopy (TSCD4D; Leica, Germany). TIFF-format images were processed through Adobe Photoshop (v.6.0; Adobe Systems, San Jose, CA). Confocal images were restored to grayscale and optimized for contrast. P2X₃ polyclonal rabbit antiserum (Neuromics, Bloomington, MN) was targeted to residues 383-397 (VEKQSTDSGAYSIGH), P2X₄ polyclonal rabbit antiserum (Alomone Laboratories, Israel) was targeted to residues 370-388 (YVEDYEQGLSGEMNQ), and P2X₂ was targeted to residues 96-113(VSIITRIEVTPSQTLGTC) (Kanjhan et al.,1996). Antibody specificity was confirmed by omission of the primary antibody and by pre-adsorption of the primary antibody with the target peptide.

The P2X_2-3 receptor cDNA was prepared by substituting an Nde1-EcoR1 DNA fragment from full-length P2X_2-1 with an Nde1-EcoR1 DNA fragment from a PCR 2.1 TA clone containing a 600-bp PCR product of the P2X_2-3 variant obtained by RT-PCR from cochlear tissue. The rat P2X₃ receptor cDNA was provided by our University College London collaborators. Full-length cDNAs for P2X_2-3 and P2X₃ were ligated into the mammalian expression vector pcDNA3.1 (Invitrogen) using BamH1-EcoR1 and Hind111-EcoR1 sites,respectively. Plasmids were linearized with EcoRV for P2X_2-3 and EcoR1 for P2X₃and cRNA synthesis was directed by T7 RNA polymerase (CapScribe; Roche). Oocytes were harvested from anesthetized Xenopus laevis frogs and defolliculated as previously described(Wildman et al., 2002). P2X_2-3 and P2X₃ cRNAs were co-injected into defolliculated Xenopus oocytes using a 1:2 ratio. Oocytes were incubated at 18°C for 48 hours and then analyzed using two-electrode voltage-clamp recording.

Agonist-activated membrane currents were recorded from oocytes injected with P2X receptor cRNAs using a two-electrode voltage-clamp amplifier(Axoclamp 2A; Axon Instruments, Union City, CA). The voltage and current electrodes were filled with 3.0 M KCl. Oocytes were superfused with Ringer's solution (pH 7.5) (containing 110 mM NaCl, 2.5 mM KCl, 1.8 mM CaCl₂and 5 mM HEPES). Agonists were prepared in normal Ringer's solution and superfused (12 ml/minute) by a gravity-fed continuous flow system allowing rapid addition and washout of compounds. Agonists were added until peak current response was reached and then washed out for 20 minutes. Data for concentration-response curves were normalized to the maximum current(I_max) evoked by ATP (100 μM). Data are presented as the mean±s.e.m. of three or more datasets using different oocytes. Concentration-response curves and inhibition curves were fitted by non-linear regression analysis (Prism v2.0, GraphPad).

ATP and related nucleotides were obtained from Sigma-Aldrich and TNP-ATP from Molecular Probes (Eugene, OR). All reagents were AnalaR grade from Sigma-Aldrich.

Whole-cell voltage-clamp of the SGN in situ in the cochlear slices was performed as described in our previous pharmacological characterization of these neurons (Salih et al.,2002). Gigaseal recordings were obtained using micropipettes with a mean input resistance of ∼1.5 mOhms. The recording electrodes were filled with 140 mM KCl, 10 mM NaCl, 2 mM MgCl₂, 5 mM HEPES, 5 mM EGTA and 10 mM glucose. ATP and associated agonists and antagonists were bath-or pressure-applied through a micropipette with the SGN voltage-clamped at -60 mV. Currents were recorded with PClamp6 software and an Axopatch 200B patch amplifier, low-pass filtered at 1-5 kHz and digitized at 5-20 kHz with a DigiData 1200 series interface (Axon Instruments, Foster City, CA). Voltage errors caused by series resistance were compensated at 70-90% online, and residual errors corrected during analysis. Junction potentials were compensated during analysis. Data are presented as mean±s.e.m.

P4 rat spiral ganglion explants were cultured as previously described(Aletsee et al., 2003). The explants were placed into tissue culture in 24-well plates (Costar) for three days using Dulbecco's modified Eagle's medium (DMEM; Invitrogen), with N-2 supplement (Invitrogen), 25 mM HEPES, 4.5 mg/ml glucose and 30 U/ml penicillin. The media was supplemented with BDNF (1 or 10 ng/ml; UpState) with or without ATPγS (100 μM; Sigma) or α,βMeATP (100 μM;Sigma). Media was changed daily. The explants were fixed with paraformaldehyde(4% in 0.1 M PBS, pH 7.3) and then immunostained for neurofilament (1:500 rabbit anti-neurofilament, 200 IgG; Sigma no. N4142), with an FITC-conjugated donkey anti-rabbit IgG (1:100; Jackson ImmunoResearch no. 711-095-152)secondary antibody. Neurites were evaluated from digital epifluorescence images of each explant obtained using the Spot imaging system (RT-color model 2.2.1 camera; Spot v.4.6 software; Diagnostic Instruments, USA). This includes a tracing algorithm for determining the length and number of neurites. Neurites were traced from their point of exit from the explant to their termination. All of the neurites on each explant were traced, along their entire length. The program computed the length of each traced neurite as the sum of vector lengths along the trace. Numbers were determined as a count of the separate traces performed on each explant image. Data were analyzed by ANOVA and Fisher's post-hoc test, with Bonferroni correction for multiple tests.

P2X receptor-activated membrane currents were recorded under voltage-clamp(V_H=-60 mV) from a subset of neonatal (P2-P4) rat SGN(n=16) to confirm their phenotype prior to extraction of the cytoplasm for RT-PCR analysis. The cells had inward current responses to focally applied ATP and α,βMeATP, a P2X₃- and P2X_2/3-selective agonist [mean=-727±88 pA (n=9) and-350±127 pA (n=7), respectively; Fig. 3A-C], consistent with our previous detailed pharmacological analysis of these neurons(Salih et al., 2002). Five neonatal rat SGN were examined for P2X_2/3 heteromeric assembly using the selective antagonist A-317491 [500 nM, after Jarvis et al.(Jarvis et al., 2002)]. Inward current responses to focal application of ATP (100 μM, 5 seconds) averaged-354±71.1 pA. Application of ATP in the presence of A-317491 produced a 47.2±7.5% block of the ATP response (P=0.019, Student's paired t-test; Fig. 3C). Recovery from the block averaged 35% after 10 minutes of antagonist washout.

The relative abundance of all seven P2X receptor transcripts was determined in SGN using single-cell real-time RT-PCR. Cytoplasm of individual SGN from the cochlear slices was aspirated into glass micropipettes. The single-cell mRNA sample was reverse transcribed into cDNA and divided into samples that provided a template for analysis of the seven P2X receptors and positive control genes. Fig. 3D shows examples of single-cell real-time PCR amplification plots from cells analyzed for P2X₃ cDNA. The C_T values assigned for each neuron sample ranged between 34 and 38 cycles, corresponding to between 3 and 159 P2X₃ cDNA molecules based on calibration curves run in tandem (see Materials and Methods). In a sample of 48 individual neurons, there was a clear difference in transcript levels across the seven P2X receptor subunits(P<0.0001; one-way ANOVA). P2X₃ and P2X₂were the most prevalent cDNAs; P2X₃ averaged approximately twice the number of transcript copies (mean=48±9 copies per cell) compared with P2X₂ (mean=25±5) (P=0.004; Student's paired t-test) (Fig. 3E). The mean P2X₄ transcript copy number detected was 17±4. The remaining P2X subunit mRNAs were detected at between two and five copies per cell (Fig. 3E). P2X₂, P2X₃ and P2X₄ transcripts were detected in most neurons [37/48 (77%), 41/48 (85%) and 40/48 (83%), respectively],whereas the detection frequency of the other P2X subtypes ranged from 17-27%. Fig. 3F shows the normalized copy number for each of the seven P2X receptor subunit isoforms expressed by individual neurons, demonstrating the dominance of the P2X₂ and P2X₃ transcripts. As a control for specificity of this single-cell gene quantification assay, transcript numbers of the housekeeping genes Gapdh and Nse were also assessed and the mRNA abundance of these targets was comparable to P2X transcript levels(Fig. 3E). The mean transcript numbers of 62±16 for Gapdh and 11±3 for Nsewere in keeping with the expected range for housekeeping genes(Warrington et al., 2000). Samples taken from bath solution and the no-template controls remained clear for the full 50 PCR cycles. No signal was obtained from neuron samples when reverse transcription was omitted (-RT control, n=5).

The relatively higher expression of P2X₂ and P2X₃mRNAs in the SGN was compatible with multimeric P2X receptor subunit assembly. Confocal immunofluorescence (Fig. 3G) confirmed the localization of P2X₂ and P2X₃ protein to the plasma membrane of the spiral ganglion cell bodies, as shown previously by our group(Huang et al., 2005; Järlebark et al., 2000). By contrast, P2X₄ immunolabeling of the spiral ganglion was considerably weaker, with diffuse signal in the soma of the SGN(Fig. 3G). When considered alongside the single-cell real-time RT-PCR analysis, these data indicate the likely dominance of an ATP-gated ion channel assembled with a P2X₃:P2X₂:P2X₃ subunit stoichiometry, as reported in a recombinant expression model(Jiang et al., 2003). However,the unique pharmacology of the SGN (Salih et al., 2002) is incompatible with the properties of the recombinant P2X_2-1/P2X₃/P2X₃heteromer (Jiang et al., 2003; Liu et al., 2001),particularly with regard to agonist profile and desensitization rate. Given the prevalence of alternative splicing of the P2X₂ (also known as P2rx2) gene in the SGN(Salih et al., 1998), the potential contribution of such splicing to the putative SGN P2X_2/3heteromer was investigated.

We have previously shown that SGN of adult rat cochlea express mRNA for three P2X₂ splice variants(Salih et al., 1998), two of which, P2X_2-1(Brake et al., 1994) and P2X_2-2(Brändle et al., 1997),have widespread distribution in other tissues and have been characterized utilizing heterologous expression systems. The third isoform,P2X_2-3(Salih et al., 1998), has a 39-bp deletion adjacent to the carboxy-terminus coding region (13 amino acid truncation). A schematic showing P2X_2-3 compared with the other two isoforms is presented in Fig. 4A. The expression profile for these three P2X₂transcripts in SGN of neonatal cochleae was determined by end-point single-cell RT-PCR. Fig. 4Bshows the predominance of the P2X_2-3 splice variant. In a sample of 31 neurons, P2X_2-3 had the highest detection frequency (74%), and most commonly, was the only P2X₂ isoform detected (Fig. 4C). This suggests that the P2X_2-3splice variant is the most likely P2X₂ isoform within a putative SGN P2X_2-3/P2X₃/P2X₃heteromer, henceforth referred to as the P2X_2-3/3 receptor.

The P2X_2-3/3 receptor was functionally characterized in the Xenopus oocyte expression system to determine its likely candidacy as the native neonatal rat SGN P2X receptor. Initial experiments confirmed the functionality of the P2X_2-3 subunit as a homomer(Table 3; D.G. and G.D.H.,unpublished). Notable characteristics included an EC₅₀ of 9 μM,and selective activation by ATP and 2MeSATP, but not α,βMeATP. Based on the relative abundance of P2X transcripts determined by real-time RT-PCR, P2X_2-3 and P2X₃ cRNAs where then co-injected into Xenopus oocytes in a 1:2 ratio. The resulting P2X_2-3/3 receptor showed an extended agonist sensitivity to include α,β-meATP and ADP(Fig. 5A). This profile does not match the previously characterized recombinant P2X_2-1/3 receptor(Liu et al., 2001), as shown by its sensitivity to 2MeSATP and ADP. In addition, this novel P2X_2-3/3 heteromer also had greater sensitivity to ATP (EC₅₀=0.4 μM at pH 7.5) than the P2X_2-1/3 heteromer(Fig. 5B; Table 3). At pH 6.5, the P2X_2-3/3 receptor exhibited enhanced sensitivity to ATP because of positive allosteric modulation by protons, as previously reported for P2X_2-1 homomers(Kanjhan et al., 2003; King et al., 1996; Stoop et al., 1997). This positive allosteric modulation of the ATP response is pronounced in the native SGN (Salih et al., 2002). Furthermore, the antagonist trinitrophenyl ATP (TNP-ATP) inhibited the recombinant P2X_2-3/3 receptor(IC₅₀=3.5 μM) similar to the SGN. The correlation in phenotype between the recombinant P2X_2-3/3receptor and the properties of the P2X receptor in the neonatal rat SGN is summarized in Table 3.

To establish the likely temporal expression profile for the P2X_2-3/3 receptor, end-point RT-PCR was performed using total RNA isolated from rat spiral ganglion tissue at postnatal ages P0-P14 (Fig. 6). This spanned the period of afferent neurite reorganization in the cochlea. These data demonstrated that at P0, all seven P2X transcripts were present. However, the P2X₁ transcript diminished after P0. P2X₃and P2X₅ transcripts were downregulated by P14. Expression of P2X₂, P2X₄, P2X₆ and P2X₇transcripts was detected at all ages. These findings provide a consolidation of analysis of P2X receptor expression in the developing rat spiral ganglion,and is supported by several preceding studies(Brändle et al., 1999; Housley et al., 1998; Järlebark et al., 2000; Nikolic et al., 2001; Nikolic et al., 2003; Salih et al., 1998). These data are compatible with our earlier analysis of P2X₃ receptor expression in the rat and mouse that demonstrated coincident gene transcription and protein translation(Huang et al., 2005; Huang et al., 2006). The downregulation of P2X₃ by P14 indicates that the putative P2X_2-3/3 receptor is specific to the period of neurite development.

The functional significance of the P2X_2-3/3 receptor on neurite growth was investigated in a neonatal rat spiral ganglion explant model. Explants from P4 rat spiral ganglia were cultured for 72 hours in the presence of BDNF(Brors et al., 2002) with and without the P2X receptor agonists ATPγS and α,βMeATP (100μM). BDNF (10 ng/ml) promoted SGN neurite extension by 57%(P<0.05, n=11; Student's unpaired t-test) and increased the neurite number by 51% (P<0.01) (compare Fig. 7D with Fig. 7A,G-J). The neurotrophic action provided by BDNF was dose-dependent, with no significant effect at 1 ng/ml (Fig. 7G,H). The ATP analogs, while having no independent effect (P>0.05; compare Fig. 7B,C with Fig. 7A,G-J), strongly inhibited the BDNF-dependent (10 ng/ml) trophic action on neurite growth and development. ATPγS eliminated the BDNF-dependent neurite extension(P<0.001; Fig. 7E,G) and reduced the BDNF-dependent increase in neurite number by∼50% (P<0.05; compare Fig. 7D with Fig. 7E,H).α,βMeATP reduced the BDNF-dependent neurite extension by 27%(P=0.002; compare Fig. 7D with Fig. 7F,I)and number by 21% (Fig. 7F,J). These data demonstrate a purinergic inhibitory control of BDNF-dependent neurotrophism in the spiral ganglion invoked by putative P2X_2-3/3 receptor activation.

Reorganization of synaptic contacts between the primary auditory neurons and the sensory hair cells of the developing rat cochlea is required to ensure that the type I fibers exclusively innervate IHC, while the neurites from the type II SGN innervate the OHC. This requires withdrawal of the type I fibers from the OHC, and pruning of their synapses with IHC, a process that is largely resolved in the first postnatal week in the rat cochlea (for a review,see Rubel and Fritzsch, 2002). Here we show that coincident with this period, there is tight gene regulation of the P2X₃ subunit and a bias towards the P2X_2-3 subunit splice variant. The match of phenotype of the recombinant P2X_2-3/3 receptor with that of the native SGN P2X receptor membrane-current properties indicates that this is the principal P2X receptor type in the SGN over this crucial period for neurite reorganization in the cochlea. The effect of the activation of this receptor type by the P2X_2/3 receptor agonists ATPγS and α,βMeATP leads to potent inhibition of the extension and branching of the SGN. These data suggest that the molecular signaling, which leads to withdrawal of neurites from inappropriate targets, may involve the release of extracellular ATP, which, acting through the P2X_2-3/3 receptor, diminishes the neurotrophic signaling between the hair cells and the SGN neurites.

As noted, P2X receptors are ATP-gated ion channels with a trimeric subunit configuration (Nicke et al.,1998), which, in the case of P2X₃/P2X₂/P2X₃ heteromers, incorporates one P2X₂ subunit and two P2X₃ subunits(Jiang et al., 2003). Khakh and Egan recently showed that these P2X-subunit trimers form a functional non-selective cation channel by utilizing both membrane-spanning domains of each subunit to line the channel pore(Khakh and Egan, 2005). Our study shows for the first time that P2X₂ and P2X₃ (also known as P2rx3) gene expression within individual neurons is coregulated so that transcript levels predispose the subsequent assembly of the translated proteins in the appropriate stoichiometry. An additional level of P2X receptor gene regulation has also been revealed in specificity of P2X₂ receptor mRNA splicing. It is likely that the majority of SGN neurons analyzed in this study were type I,given their preponderance; however, we have previously demonstrated that type I and type II SGN express comparable ATP-gated currents(Jagger and Housley, 2003) and both inner radial fibers (type I) and outer spiral fibers (type II) are immunopositive for P2X₂ and P2X₃. Although ATP-release from the organ of Corti has been confirmed(Wangemann, 1996), the source of this P2X receptor agonist remains unknown, but may include hair cells,given the corelease of ATP with other transmitters such as glutamate(Burnstock, 2004). Thus, it is likely that the P2X_2-3/3 receptor-mediated regulation of neurite development demonstrated here would apply to both types of SGN.

Our single-cell real-time RT-PCR study resolved significant differences in transcript copy number between the seven different P2X subunits and also between the two housekeeping genes Gapdh and Nse. The mRNA abundance for these genes ranged from <10 to hundreds of copies. This is a relative transcript measurement, detecting a proportion of mRNAs available from each neuron. The remarkable feature of the molecular analysis provided by the TaqMan-based RT-PCR is that between-sample variation is sufficiently small to enable statistically significant discrimination when single-cell transcript copy numbers vary by tens of copies (Fig. 3E). The single-cell transcript levels were comparable to those previously described for other ion channel subunits(Sucher et al., 2000) and for transcription factors and housekeeping genes(Wagatsuma et al., 2005; Warrington et al., 2000). Our application of TaqMan-based real-time PCR in single cells is also supported by recent investigations of dopaminergic receptor mRNA transcript levels(Liss et al., 2001), mGluR in rod photoreceptors (Kamphuis et al.,2003) and 5HT receptors in hypoglossal motoneurons(Zhan et al., 2002). These studies, alongside critical reviews of quantitative RT-PCR technology(Bustin, 2002; Fink et al., 1998; Ginzinger, 2002; Pfaffl et al., 2002; Wagatsuma et al., 2005),establish criteria for determining the validity of quantifying transcripts at low copy number. The key requirements are to confirm amplicon specificity, PCR efficiency and detection sensitivity. In the present study, possible amplification of genomic targets was precluded by the absence of amplicons from -RT controls. In addition, the separation of the single-cell template into 10 samples effectively eliminated potential genomic DNA targets(Johansen et al., 1995). Amplification of cDNA templates derived from sources other than individual SGN were deemed unlikely, based on the lack of amplification of no-template controls, including perfusion media.

Our data suggest that in the early postnatal SGN, the P2X_2-3 subunit is the preferentially expressed P2X₂ variant. This isoform has a 39-bp deletion resulting in a loss of 13 amino acids immediately before the cytosolic carboxy terminus(Salih et al., 1998). The identification of functionality of this PX_2-3splice variant within a P2X_2/3 heteromeric receptor complex provides further insight into the contribution of the C-terminal domain of these subunits as a regulator of agonist profile, receptor desensitization and subunit interaction in P2X receptors (for a review, see Khakh and North, 2006). The recombinant P2X_2-3 homomer had many similarities with the P2X_2-1 homomeric receptor(Brake et al., 1994)(Table 3), however, the P2X_2-3/3 heteromer produced significant changes in receptor phenotype from that reported for the P2X_2-1/3 receptor(Lewis et al., 1995),including faster desensitization, and de novo sensitivity to 2MeSATP and ADP. Studies of the other functional variant in rat,P2X_2-2(Brändle et al., 1997), as well as engineered C-terminal region variants, demonstrate influences on trafficking of the subunits, interaction with other receptor subunits,desensitization rate and ion permeability(Brändle et al., 1997; Chaumont et al., 2004; Eickhorst et al., 2002; Gendreau et al., 2003).

Evidence is emerging that P2X receptor-based signaling has neurotrophic actions. In several neuronal culture models, activation of P2X receptors complements neurotrophin activity. For example, in PC12 cells, enhanced neurite initiation is induced by application of ATP and ATPγS, at suboptimal nerve growth factor (NGF) levels(D'Ambrosi et al., 2001). NGF and ATP induced an upregulation of several P2X receptors, including P2X₂, P2X₃ and P2X₄. Given that P2X receptors have the highest Ca²⁺ permeability of transmitter-gated cation channels (Egan and Khakh,2004), it is likely that ATP, acting as a transmitter, or paracrine signaling factor has a neurotrophic action through Ca²⁺signaling (Hegarty et al.,1997). P2X receptors are also likely to complement neurotransmitter signaling elements such as glutamate and acetylcholine(Fu, 1995; Gu and MacDermott, 1997; Jo and Schlichter, 1999) to engage neurotrophic mechanisms, such as the stabilization of synapses. P2X receptors may exert conflicting influences on neural growth. In a neural tube explant model, P2X₃ receptors, activated by α,βMeATP,have been implicated in the inhibition of motor axon outgrowth during embryonic neurogenesis, affecting both neurite length and number(Cheung et al., 2005). The influence of P2X₃ receptor signaling on embryonic neurogenesis in the CNS is supported by the wide distribution of this receptor during development of the CNS, with rapid downregulation in the postnatal period(Cheung and Burnstock,2002).

In the developing cochlea, neurite growth is supported by both BDNF-TrkB and NT3-TrkC signaling pathways. Evidence suggests that release of these neurotrophins from the sensorineural tissue, including the hair cells,supports SGN development. In explant models, SGN extension and branching is greatly enhanced by treatment with these neurotrophins(Aletsee et al., 2001; McGuinness and Shepherd, 2005; Ryan et al., 2006). Our data,showing the ATPγS and α,βMe ATP-inhibition of SGN neurite development induced by BDNF, strongly supports a role for the P2X receptors in the regulation of neurite-hair cell trophism. Developmental regulation of P2X₃ receptor expression to the period when the neurites are reorientated into the mature configuration, and a bias towards P2X_2-3 expression, all combine to suggest that ATP provides a signal through a specific P2X receptor -P2X_2-3/3. This novel receptor acts to inhibit BDNF neurotrophism and may provide the signal that induces disengagement of synaptic connections in the early postnatal rat cochlea, as a prelude to the final neural differentiation required for functional auditory neurotransmission.

We thank Joanna Stewart for assistance with statistical analysis and the University of Auckland Biomedical Imaging Research Unit for assistance with confocal microscopy. Supported by the Health Research Council (NZ), Marsden Fund (Royal Society NZ) and the James Cook Fellowship (G.D.H., Royal Society NZ).