Abstract
Parasympathetic neurons do not require neurotrophins for survival and are thought to lack high-affinity neurotrophin receptors (i.e., trks). We report here, however, that mRNAs encoding both brain-derived neurotrophic factor (BDNF) and its high-affinity receptor tropomyosin-related kinase B (trkB) are expressed in the parasympathetic chick ciliary ganglion (CG) and that BDNF-like protein is present in the ganglion and in the iris, an important peripheral target of ciliary neurons. Moreover, CG neurons express surface trkB and exogenous BDNF not only initiates trk-dependent signaling, but also alters nicotinic acetylcholine receptor (nAChR) expression and synaptic transmission. In particular, BDNF applied to CG neurons rapidly activates cAMP-dependent response element-binding protein (CREB), and over the long-term selectively upregulates expression of α7-subunit-containing, homomeric nAChRs (α7-nAChRs), increasing α7-subunit mRNA levels, α7-nAChR surface sites, and α7-nAChR-mediated whole-cell currents. At nicotinic synapses formed on CG neurons in culture, brief and long-term BDNF treatments also increase the frequency of spontaneous EPSCs, most of which are mediated by heteromeric nAChRs containing α3, α5, β4, and β2 subunits (α3*-nAChRs) with a minor contribution from α7-nAChRs. Our findings demonstrate unexpected roles for BDNF-induced, trk-dependent signaling in CG neurons, both in regulating expression of α7-nAChRs and in enhancing transmission at α3*-nAChR-mediated synapses. The presence of BDNF-like protein in CG and iris target coupled with that of functional trkB on CG neurons raise the possibility that signals generated by endogenous BDNF similarly influence α7-nAChRs and nicotinic synapses in vivo.
- ciliary ganglion
- nicotinic acetylcholine receptor
- BDNF
- trkB
- patch-clamp
- neurotrophin
- bungarotoxin
- EPSC
- cAMP response element-binding protein (CREB)
- PACAP
Introduction
Neurotrophins [nerve growth factor (NGF), brain-derived neurotrophic factor (BDNF), and neurotrophin-3 (NT-3)] act via high-affinity tyrosine kinase-containing receptors (trkA, trkB, and trkC, respectively) to support the survival and growth of diverse neuron populations and influence the form and function of chemical synapses (Lewin and Barde, 1996; Kaplan and Miller, 2000; Huang and Reichardt, 2001). In particular, BDNF and sometimes NT-3, exert rapid, primarily presynaptic effects at central, autonomic, and neuromuscular synapses and produce long-term presynaptic and postsynaptic changes consistent with altered gene expression (for review, see Lewin and Barde, 1996; Schuman, 1999; Poo, 2001). Thus, in addition to providing trophic support, neurotrophins also induce trk-dependent acute and long-term changes that coordinately influence synaptic interactions.
Parasympathetic neurons typified by those in the chicken ciliary ganglion (CG) do not require neurotrophins for survival (Helfand et al., 1976; Rohrer and Sommer, 1982; Lindsay et al., 1985; Krieglstein et al., 1998). Instead, CG neurons rely on other growth factors, notably ciliary neurotrophic factor (CNTF) (Leung et al., 1992; Finn et al., 1998) and glial-derived neurotrophic factor (GDNF) (Hashino et al., 2001) for trophic support. Moreover, studies using Northern and RNase protection assays failed to detect trk mRNA in ciliary ganglia (Dechant et al., 1993; Hallbook et al., 1995). These observations have led to the presumption that CG neurons lack trks (Huang and Reichardt, 2001).
As with sympathetic ganglion neurons and skeletal muscle fibers, fast chemical synapses on ciliary and other parasympathetic ganglion neurons are mediated by nicotinic acetylcholine receptors (nAChRs). In sympathetic neurons, NGF supports the expression ofα3-nAChR subunit protein (Yeh et al., 2001), an effect mirrored in pheochromocytoma cell line (PC12) cells, where NGF increases α3, α5, α7, β2, and β4 nAChR subunit mRNAs as well as nAChR function (Henderson et al., 1994; Takahashi et al., 1999). Also, sympathetic neurons overexpressing BDNF display increased preganglionic innervation density (Causing et al., 1997), indicative of long-term presynaptic effects. BDNF acting through trkB also regulates neuromuscular junction form and function. For example, BDNF rapidly enhances presynaptic release to increase the frequency and amplitude of spontaneous nAChR-mediated synaptic currents in nerve–muscle cultures (Lohof et al., 1993; Stoop and Poo, 1996). Over the long term, BDNF restores neuregulin levels, restricts axon sprouting, and maintains postsynaptic architecture in muscle disrupted by activity blockade (Loeb et al., 2002), whereas sustained trkB-mediated signaling is likely required to maintain postsynaptic nAChR clusters (Gonzalez et al., 1999). These findings prompted us to speculate that previous assays were perhaps insufficiently sensitive to detect trks expressed in ciliary ganglia and that neurotrophin–trk signaling, although not required for trophic support, might influence the components and function of nAChR-mediated synapses on CG neurons. We focused on BDNF–trkB signaling, and we have demonstrated expression of BDNF-like protein in ciliary ganglia and functional trkB on CG neurons. To explore synaptic relevance, the impact of BDNF–trkB signaling on α7-nAChRs and α3*–nAChR-mediated synapses was assessed using CG neurons grown in cell culture. BDNF treatment upregulated expression of α7-nAChRs after several days and increased the frequency of spontaneous synaptic currents within minutes. The results reveal an unanticipated relevance for BDNF–trkB signaling in parasympathetic CG neurons.
Materials and Methods
Neurons. CG neuron cultures were prepared under sterile conditions from embryonic day 8 (E8) chick embryos. Dissociated neurons were plated at one or two ganglion equivalents in 15 mm diameter polystyrene tissue culture wells or on 12-mm-diameter glass coverslips; both substrates were precoated with poly-dl-ornithine and laminin (Pugh and Margiotta, 2000; Chen et al., 2001). The standard culture medium consisted of minimum essential medium (MEM) containing 100 U/ml penicillin, 100 μg/ml streptomycin, 2 mm glutamine, and 10% heat inactivated horse serum (MEMhs; all components from Invitrogen, Rockville, MD) and was supplemented with 3% embryonic eye extract (Nishi and Berg, 1981). Neurons were maintained at 37°C in 95% air and 5% CO2 for 4–7 d and received fresh culture medium every 2–3 d, conditions that support 100% survival of CG neurons for at least 7 d (Nishi and Berg, 1981). In test cultures the medium was further supplemented with BDNF (50 ng/ml, unless indicated otherwise) sometimes in conjunction with other reagents as described for individual experiments in Results. For some studies, CG neurons were acutely dissociated from E8 or E14 ganglia, as previously described (McNerney et al., 2000; Nai et al., 2003). Neurons were plated on acid-washed, poly-d-lysine-coated glass coverslips in electrophysiological recording solution (RS) containing (in mm): 145.0 NaCl, 5.3 KCl, 5.4 CaCl2, 0.8 MgSO4, 5.6 glucose, and 5.0 HEPES, pH 7.4 (Dichter and Fischbach, 1977) that was supplemented with 10% heat-inactivated horse serum (RShs). Acutely dissociated neurons were maintained in RShs at 37°C for 2–4 hr before use.
Conventional RT- PCR. The presence of mRNA encoding chicken trkB, BDNF, α7, and α3-nAChR subunits, as well as β-actin (βA) or glyceraldehyde-3-phosphate dehydrogenase (GAPDH), was assessed by conventional reverse transcriptase-based PCR (RT-PCR) as previously described (Burns et al., 1997). Briefly, RNA was isolated from E8–E15 chick tissues or from CG neuron cultures using a one-step kit (RNAqueous; Ambion, Austin, TX). Total tissue RNA (1 μg) was treated with Amplification Grade RNase-free DNase (1 U at 1 U/μl; Invitrogen), and then 25–200 ng of DNase-treated RNA used to synthesize cDNA using Superscript II reverse transcriptase (RT+; Invitrogen). The resulting cDNAs were then used as templates for PCR amplifications in 25 μl reaction volumes containing 50 mm KCl, 20 mm Tris-HCl, 2.5 mm MgCl2, 200 μm dNTPs, 5 U/μl TaqDNA polymerase (Invitrogen), and 0.4 μm forward (F) and reverse (R) oligonucleotide primers (synthesized by Marshall University DNA Core Facility, Huntington, WV). The chicken-specific primers used were: trkB (Dechant et al., 1993): F, C1156TTCAGCTGGACAACCCTAC1175; RK+, T1868GGAAGTCCTTGCGGGCATT1849; RK–, GCCCCTCTCTCATCTT; BDNF (Maisonpierre et al., 1992): F, G287CAGTCAAGTGCCTTTG303; R, G748AGCCCACTATCTTCCCC731; α7-nAChR subunit (Couturier et al., 1990b): F, G1092GGGAAAAATGCCTAAAT1109; R, G1614ACAGCCTCTACAAAGTT1597; α3-nAChR subunit (Couturier et al., 1990a): F, A985TGCCTGTATGGGTGAGAACT1005; R, T1226TGCCACTGAAATCGGAAAAC1206; GAPDH (Stone et al., 1985): F, G532CCATCACAGCCACACAGAA551; R, A980CCATCAAGTCCACAACACG961; and β-actin (GenBank accession number L08165): F, A860TCTTTCTTGGGTATGGA877;R,A1134CATCTGCTGGAAGGTCC1117.
The two trkB primer pairs (F/RK+ and F/RK-) correspond to those shown previously to amplify kinase-containing (full-length) and truncated (kinase-deleted) chicken trkB isoforms, respectively (Garner et al., 1996). The F/RK+ pair is not predicted to hybridize with chicken trkA (Schropel et al., 1995) or trkC (Garner and Large, 1994) cDNAs. The α7- and α3-nAChR subunit primer pairs both amplify products within nonconserved regions of their respective cytoplasmic domains, located between transmembrane segments III and IV (Schoepfer et al., 1990). The trkB and AChR subunit primers were optimized for amplification, and the reactions were performed in the linear range of the assay (25–29 cycles). PCR products were separated on 1.0% agarose gels stained with ethidium bromide. Identical reactions lacking RT served as controls for possible amplification of genomic DNA and were consistently negative. Changes in the levels of α7 and α3 mRNAs in response to BDNF treatment were estimated semiquantitatively after digitizing gel images using Kodak 1D Image Analysis software (Eastman Kodak, Rochester, NY) from the ratio of PCR product intensities to those of βA from the same cultures.
Real-time PCR. Changes in α7- and α3-nAChR subunit mRNA levels induced by BDNF were confirmed using RT-based real-time PCR. cDNA samples corresponding to 50 ng of input RNA were combined with Taqman universal PCR master mix (Roche, Branchburg, NJ), F and R primers (0.4 μm), and Taqman probe (0.1 μm) [with 6-FAM (carboxyfluorescein, reporter dye) and TAMRA (tetramethylrhodamine, quencher dye) inserted at 5′ and 3′ ends, respectively]. Selection of the following primers and probes was optimized using Applied Biosystems (Foster City, CA) Primer-Express software, with α7- and α3-nAChR subunit primers chosen to amplify regions within transmembrane segments III and IV, and span intron-exon boundaries (Schoepfer et al., 1990): α7-nAChR subunit (Couturier et al., 1990b): F, C1020CATGATTATTGTTGGCCTCTCT1042; R, T1210CGGCCCTGTTTATGTTGAC1190; Probe, A1115GAGTCATCCTTCTGAATTGGTGTGCTTGGT1145; α3-nAChR subunit (Couturier et al., 1990a): F, G1178CAGCTGCTGCCAGTACCA1196; R, A1398ATGACCATGGCAACATATTTCC1376; Probe, T1216TCAGTGGCAATCTCACAAGAAGTTCCAGC1245; and GAPDH (Stone et al., 1985): F, C1795CGTCCTCTCTGGCAAAGTC1814; R, A2374ACATACTCAGCACCTGCATCTG2352; Probe, A2211TCAATGGGCACGCCATCACTATCTTCC2228.
Twenty-five microliter PCRs were performed in triplicate using a GeneAmp 5700 sequence detection system (Applied Biosystems). This system allows the increase in PCR product to be monitored directly based on the threshold number of cycles (CT) required to produce a detectable change in fluorescence (ΔF) resulting from the release of probe. Relative levels of α7- and α3-nAChR cDNA (Rα7, Rα3) in control and BDNF-treated cultures were calculated from the difference in CT values (ΔCT = CTcontrol – CTBDNF) for α7 or α3 amplifications (ΔCTα7, ΔCTα3) compared with those for the housekeeping gene, GAPDH (ΔCTGAPDH) using: (1) In Equation 1, Eα and EGAPDH are the real-time PCR amplification efficiencies determined in separate studies from the slope of CT versus input log cDNA dilution, where E = 10-1/slope. E values for amplifying α7, α3, and GAPDH cDNAs were 2.10, 2.10, and 2.23, respectively.
Immunocytochemistry. A polyclonal antibody (pAb) generated against the extracellular domain of chicken trkB (#R22781) that does not recognize trkA or trkC (von Bartheld et al., 1996) was generously provided by Dr. Frances Lefcort (Montana State University). pAb recognizing Ser133-phosphorylated cAMP response element (CRE) binding protein (p-CREB) was purchased from Cell Signaling Technology (Beverly, MA). Ciliary and dorsal root ganglia (DRG) were fixed for 1–4 hr in 4% paraformaldehyde prepared in 0.15 m PBS at pH 7.4 (PBS), washed in PBS, cryoprotected in PBS containing 30% sucrose, embedded in OCT (Miles Laboratories, Elkhardt, IN), cryosectioned at 10 μm, and mounted on glass slides. After rehydration, sections were blocked for 1 hr at room temperature in 30 mm Tris and 150 mm NaCl containing 0.4% Triton X-100, 1% glycine, 10% goat serum, and 3% bovine serum albumin. trkB antibody (Ab) was applied to sections in blocking solution containing 4% goat serum (1:1000, 4°C, 16 hr), and after washing, secondary Ab (AlexaFluor594-conjugated anti-rabbit IgG; Molecular Probes, Eugene, OR) was applied in the same solution (1:400, 22°C, 1 hr). Sections were then washed, dipped in distilled water, and mounted in Vectashield (Vector Laboratories, Burlingame, CA). Acutely dissociated CG neurons or CG neuron cultures, both on glass coverslips, were fixed for 0.5–1.0 hr in 2–4% paraformaldehyde and blocked in PBS containing 10% donkey or goat serum. Coverslips were then incubated in trkB Ab (1:2000, 37°C, 2 hr) treated with Cy3-conjugated anti-rabbit IgG (1:400, 1 hr, 37°C; The Jackson Laboratory, Bar Harbor, ME) in PBS containing 4% serum, washed, and mounted. A similar protocol was followed for p-CREB immunostaining except that the block and wash buffers contained 0.1% Triton X-100, p-CREB Ab was applied (1:400, 4°C, 16 hr), and the secondary Ab was AlexaFluor488-conjugated anti-rabbit IgG (1:400, 22°C, 1 hr).
Image analysis. Immunostained preparations were viewed using epifluorescence microscopy (BX50, UplanFL 40×, 0.75 numerical aperture objective; Olympus, Tokyo, Japan), and images were acquired and processed using a SenSys KAF-1400 cooled digital CCD camera under the control of IP Lab software (version 3.6; Scanalytics, Reading, PA) as described previously (Chen et al., 2001). Neurons were considered p-CREB-positive if the mean fluorescence intensity of pixels in an elliptical region of interest (ROI) superimposed over the nucleus exceeded that of the ROI when placed over cytoplasm by >15%.
ELISAs. The presence of BDNF in chicken tissue homogenates and tissue culture medium components was assessed using a commercial BDNF sandwich ELISA kit having no significant cross-reactivity with NGF, NT4/5, or NT3 (Chemikine; Chemicon, Temecula, CA). The ELISA uses rabbit polyclonal antibodies (raised against human BDNF) to capture BDNF from the sample, and a biotinylated mouse monoclonal antibody to detect the captured BDNF. Because mammalian and chicken BDNF share all but seven amino acids, with the mismatches distributed along the entire length of the peptide (Isackson et al., 1991), the kit antibodies likely recognize chicken BDNF. Nevertheless, we refer here to detection of “BDNF-like protein,” with levels quantified within the linear range of the assay (7.8–500 pg/ml) using recombinant human BDNF as standard.
α-Bungarotoxin binding. CG neurons were plated at one or two ganglion equivalents per well and grown in culture wells for 4–5 d. Neurons in triplicate culture wells were washed twice in MEMhs, incubated in MEMhs containing 10 nm [125I]-α-Bungarotoxin (α-Bgt) (specific activity = 130–140 Ci/mmol; PerkinElmer, Boston, MA) for 1 hr at 37°C, and then washed three times with MEMhs. We previously showed that these conditions are sufficient to saturate surface α-Bgt sites on dissociated CG neurons (McNerney et al., 2000). Nonspecific binding was determined in parallel wells by including 100 μm d-tubocurarine with 10 nm [125I]-α-Bgt. After labeling and washing, the wells were scraped in 500 μl 0.6 N NaOH, the solution was collected, and [125I]-α-Bgt radioactivity was determined using a Beckman Instruments (Fullerton, CA) G-5500 gamma counter.
Electrophysiology. Whole-cell recordings were obtained at 21–23°C from CG neurons after 3–5 d in culture. Patch pipettes were fabricated from Corning 8161 glass tubing (WPI, Sarasota, FL), filled with (in mm) 145.6 CsCl, 1.2 CaCl2, 2.0 EGTA, 15.4 glucose, and 5.0 Na-HEPES, pH 7.3, and had tip impedances of 2–3 MΩ. To induce nAChR currents, neurons were bathed in RShs, held at –70 mV, and 20 μm nicotine (Nic) applied in RS by rapid pressure microperfusion (at 10–12 psi) from a delivery pipette (4–6 μm tip diameter) positioned ≈5–10 μm from the neuron soma. We previously showed that fast-onset, rapidly desensitizing α7-nAChR-mediated whole-cell currents induced by 20 μm Nic in this manner are indistinguishable in amplitude from those obtained using fast piezoelectric switching (Nai et al., 2003). The fast (α7-nAChR-mediated) and slower (α3*-nAChR-mediated) decaying current components induced by 20 μm Nic were identified and analyzed using Clampfit (pClamp 6.0 or 8.0; Axon Instruments, Foster City, CA) as previously described (Nai et al., 2003). For analysis, peak Nic-induced response component amplitudes (in picoamperes) were normalized to neuron soma membrane capacitance (in picofarads). To quantify BDNF effects, whole-cell Nic responses (in picoamperes per picofarad) obtained from treated neurons were normalized to those for control neurons from the same cultures. To assess synaptic function, spontaneous EPSCs (sEPSCs) were acquired at –70 mV for 2–5 min, without stimulation, as previously described (Chen et al., 2001). For these experiments, horse serum was sometimes omitted from the recording solution, without discernible effect on the results. Synaptic current frequency and amplitude analyses were subsequently performed using either BASIC-23 programs written in-house, or commercially available software (Mini Analysis 5.6.12; Synaptsoft Inc., Decatur, GA). Briefly, sEPSC frequency values obtained from BDNF-treated neurons were normalized to those from control neurons from the same culture platings. In addition, sEPSCs were extracted from selected records displaying >50 non-overlapping events, and the component amplitude and decay time constant values were pooled for control and BDNF-treated neurons.
Statistics. All parameter values are expressed as mean ± SEM. Unless indicated otherwise, the statistical significance of paired and unpaired numerical comparisons was determined using the appropriate two-tailed t test (p < 0.05).
Results
Expression of trkB mRNA and protein
PCR primers specific for kinase-containing (K+) full-length trkB (Garner et al., 1996) amplified a ≈700 bp product from both E8 and E14 CG cDNA templates (Fig. 1a,b). The CG product size was consistent with that predicted for chicken trkB (713 bp) (Garner et al., 1996) and indistinguishable from that obtained in amplifications from E15 DRG, previously shown to express abundant trkB mRNA (Hallbook et al., 1995) and protein (Anderson, 1999; Rifkin et al., 2000). In addition, trkB products from E14 CG and E15 DRG yielded identical restriction profiles after digestion with BamHI (data not shown) or HbaII (Fig. 1c), with fragment sizes as predicted for digestion of K+ trkB cDNA (Dechant et al., 1993). Truncated trkB isoforms lacking the kinase domain (K–) but containing variable juxtamembrane insertions are also expressed in the chicken nervous system (Garner et al., 1996), and K–-specific primers amplified products of expected sizes (≈ 400, 500, and 600 bp) from both DRG and CG (Fig. 1a,b). In each case, the PCR amplifications from DRG and CG sources were specific for cDNA in the sense that they were absent when the synthesis reaction lacked RT (data not shown). Although the significance of the truncated trkB transcripts was not studied here, the results demonstrate that both truncated and full-length trkB transcripts are expressed in CG during E8–E14, a developmental window when nicotinic synapses formed on the neurons undergo substantial structural and functional maturation (Landmesser and Pilar, 1972, 1974a).
The presence of trkB protein on CG neurons was demonstrated by fluorescence immunolabeling (Fig. 2) using an Ab that recognizes the extracellular domain of chicken trkB (but not trkA or trkC; von Bartheld et al., 1996). Specific trkB labeling, similar to but somewhat less intense than that seen for E15 DRG sections, was evident in both E8 and E14 CG sections (Fig. 2a,b,c,f) and was localized to the neuron surface where it increased between the two developmental ages. In CG neuron cultures, the somata and processes of neurons displayed specific trkB labeling that became more extensive and intense between 8 hr and 4 d in culture (Fig. 2d,e,g), a period when functional synapses are formed and increase in activity (Chen et al., 2001). At 4 d in culture, ∼80% of CG neurons scored positive for trkB immunoreactivity. These findings demonstrate that trkB protein is expressed by CG neurons and, because dissociated neurons in acute and culture preparations were not permeabilized, indicate that a substantial fraction is localized on the cell surface. Taken together, the mRNA and protein studies further suggest that catalytically competent, high-affinity BDNF receptors are present in the ganglion and that their expression on CG neurons increases during periods of synaptic differentiation both in vivo and in cell culture.
Functional relevance of BDNF and trkB
To be relevant in vivo, endogenous BDNF should both be present in the CG and be able to elicit trk-dependent signaling in the neurons. Because BDNF detected in spinal cord ventral horn results from both local synthesis and retrograde transport to motor neurons from striated muscle (Koliastsos et al., 1993), we tested for the presence of BDNF both in CG and in the iris, a ciliary neuron target that like the ciliary body is primarily striated muscle in birds (Marwitt et al., 1971). Using a commercial ELISA, we detected BDNF-like protein in E14 iris muscle as well as in E14 and in E8 CG (Fig. 3a). The assay failed to detect BDNF in 10% heat inactivated horse serum or whole eye extract, routinely used at 10 and 3%, respectively, as supplements to CG culture medium. The assay also failed to detect BDNF-like protein in intact chicken serum. Relevant to our culture experiments, we presume that dilution of BDNF derived from the iris and ciliary muscle during eye extract preparation reduces its concentration below the detection limit of the assay (7.8 pg/ml). BDNF-like protein present in E8 and E14 CG may be the source of the strong BDNF immunoreactivity previously reported at the same developmental ages for accessory oculomotor neurons (Steljes et al., 1999) which provide preganglionic input to the CG. In addition to arriving by retrograde transport from the intraocular muscle targets, however, the BDNF-like protein present in the CG may also result from local synthesis, because BDNF mRNA was detectable by RT-PCR in E8 and E14 ganglia and in CG neurons maintained in standard culture medium for 4 d (Fig. 3b).
To determine if the trkB protein expressed on CG neurons represents functional receptor, we tested the ability of applied BDNF to cause phosphorylation of CREB, a cAMP- and Ca2+-regulated transcription factor (for review, see Impey et al., 1996; Greenberg and Ziff, 2001; Deisseroth et al., 2003), whose activation is a hallmark of trk-dependent neurotrophin signaling (Finkbeiner et al., 1997) (Fig. 4). For this purpose, CG neurons grown in culture for 4–5 d were challenged with BDNF (100–200 ng/ml; 10–15 min) or, as a positive control, with pituitary adenylate cyclase activating polypeptide (PACAP; 100 nm, 10–15 min), previously shown to cause robust increases in intracellular cAMP and Ca2+ (Margiotta and Pardi, 1995; Pardi and Margiotta, 1999), and then tested the cultures for p-CREB immunoreactivity. A similar immunocytochemical approach was previously shown to provide a convenient all-or-none assay for CREB activation in single hippocampal neurons (Hu et al., 2002). After treatment with BDNF, 49 ± 6% of 321 CG neurons from 18 fields (N = 321, 18) scored as p-CREB-positive, compared with 99 ± 2% (N = 132, 9) after PACAP treatment and 3 ± 2% (N = 179, 10) in untreated control cultures assayed in parallel (Fig. 4a–d, g). Consistent with a requirement for trkB signaling, the proportion of p-CREB positive neurons induced by BDNF dropped to control levels (1 ± 1%; N = 114, 6) after 1 hr preincubation and 15 min cotreatment with K252a (Fig. 4e,g), a trk-selective tyrosine kinase inhibitor (Pizzorusso et al., 2000). Preincubation (1 hr) and 15 min cotreatment with PD98059, an MEK1 inhibitor that blocks the neurotrophin-activated, RAS-dependent signaling pathway leading to CREB activation (Ying et al., 2002) similarly reduced the proportion of p-CREB positive neurons induced by BDNF to 3 ± 2% (N = 85, 6) (Fig. 4f,g). The observation that 51% of neurons showed no detectable response to BDNF in this assay cannot be explained by limited CREB availability because nearly all nuclei were immunoreactive after PACAP treatment. The difference might instead reflect suboptimal BDNF treatment times or heterogeneity of functional trkB expression levels. In either case, the protein and p-CREB assays (Figs. 3, 4) demonstrate that an endogenous source of BDNF-like protein is present in the parasympathetic CG, where it is poised to activate functionally competent trkB receptors present on the neurons and thereby recruit appropriate signal pathways leading to CREB activation.
BDNF upregulates α7-nAChRs
Because neurotrophins are not required for full survival of CG neurons, we examined possible roles for BDNF–trkB signaling in regulating the components and function of nicotinic synapses on the neurons. α7-nAChRs were assessed as potential targets of BDNF–trkB signaling because they can rapidly modulate transmission (McGehee et al., 1995; Gray et al., 1996; Wonnacott, 1997; Margiotta and Pugh, 2004), an action resembling the increased synaptic efficacy produced by BDNF (for review, see McAllister et al., 1999; Schnider and Poo, 2000; Poo, 2001). α-Bgt binds with high affinity to α7-nAChRs (Couturier et al., 1990b) and [125I]α-Bgt was therefore used to quantify surface α7-nAChRs on CG neurons (Fig. 5a), as previously described (McNerney et al., 2000). In CG neuron cultures grown with BDNF for varying times before assay at 5 d, [125I]-α-Bgt binding was unchanged relative to control cultures after 4 hr exposure, but increased nominally (24%) after 16 hr. Extending the treatment time to the full 5 d culture period resulted in levels of α-Bgt binding that were significantly higher (62 ± 10%; p < 0.01) in CG cultures exposed to BDNF relative to control cultures assayed in parallel (N = 10 for both). Actual levels of [125I]-α-Bgt bound in control cultures and cultures treated with BDNF for 5 d were 3.9 ± 0.2 and 6.0 ± 0.4 fmol/CG equivalent, respectively. In principle, the increased levels of α7-nAChRs seen after exposure to BDNF could have been caused by activation of either trkB or the low-affinity neurotrophin receptor (p75NTR) that is also present on CG neurons (Lee et al., 2002). The latter possibility is unlikely, however, because an identical elevation (62 ± 14%; N = 3) was observed when BDNF was coapplied for 5 d with ChEX, a pAb that recognizes and blocks chicken p75NTR but not trk function (Wescamp and Reichardt, 1991). The ability of long-term BDNF exposure to upregulate α7-nAChRs may reflect increased α7-nAChR subunit gene expression because levels of α7-nAChR subunit relative to βA mRNA, determined by semiquantitative RT-PCR, were elevated significantly (by 53 ± 10%) in cultures treated with BDNF for 4–5 d compared with untreated control cultures, tested in parallel (N = 6 each) (Fig. 5b). As with the protein assays, BDNF also increased α7-nAChR subunit mRNA in cultures with p75NTR blocked by coapplication with ChEX (41 ± 15%; N = 3). Using real-time PCR, a 98 ± 26% (N = 5) increase in α7-nAChR subunit relative to GAPDH mRNA was observed (Fig. 5c). In both cases, the BDNF-induced increases in α7-nAChR subunit mRNA were selective in the sense that identical treatments failed to significantly alter α3-nAChR subunit mRNA levels (Fig. 5b,c).
Having demonstrated that BDNF induces trk-dependent increases in levels of α7-nAChR protein and mRNA in CG neurons, we next sought to determine if the same treatments also enhanced α7-nAChR-mediated currents. Rapid application of nicotine (Nic; 20 μm) to CG neurons grown in culture typically induces a whole-cell current response featuring an initial fast-desensitizing component that is blocked by α-Bgt (Fig. 6a) and hence mediated by α7-nAChRs (Pardi and Margiotta, 1999; McNerney et al., 2000; Nai et al., 2003). Whereas 65 ± 5% (N = 80 neurons; 5 platings) of CG neurons grown in standard culture medium displayed rapidly decaying, α7-nAChR-mediated currents, BDNF treatment for 3–4 d increased the proportion to 83 ± 4% (N = 76, 5). In such cases, the peak α7-nAChR current values relative to membrane capacitance (Ifast/Cm, pA/pF) were 49 ± 14% (N = 63, 5) larger for neurons from cultures treated with BDNF compared with untreated controls (N = 52, 5) tested in parallel (Fig. 6b,c). Consistent with the time course for upregulation of surface α7-nAChRs seen in the [125I]-α-Bgt binding studies, BDNF treatments for 10–30 min or 16–24 hr produced only nominal increase in Ifast/Cm relative to untreated controls tested in parallel (Fig. 6b) (data not shown, p > 0.05 for both). The slowly decaying component of the Nic-induced current is mediated primarily by heteromeric α3*-nAChRs (Nai et al., 2003) that contain α3, α5, β4, and occasionally β2 subunits, but lack α7 subunits (Vernallis et al., 1993; Conroy and Berg, 1995) and are insensitive to α-Bgt (Fig. 6a). The ability of BDNF to increase Ifast/Cm was specific for α7-nAChRs because slow currents (Islow/Cm) attributable to α3*-nAChRs and present in all neurons, were unchanged after exposure to BDNF for 10–30 min, 16–24 hr, or 4–5 d (Fig. 6b). In addition, the 4–5 d BDNF treatments had no discernible effects on membrane capacitance or the voltage sensitivity or maximal values of voltage-activated Na+ or Ca2+ currents (data not shown). In summary, the size, latency, and specificity of the increased α7-nAChR current responses seen after chronic BDNF treatment are consistent with the BDNF-activated trkB-dependent upregulation of α7-nAChR mRNA and protein that occur over a similar time course.
BDNF increases activity at nicotinic synapses
Functional synapses form between CG neurons in culture (Margiotta and Berg, 1982) and display spontaneous, impulse-driven nicotinic EPSCs (sEPSCs) (Fig. 7). We previously demonstrated that although α7-nAChRs contribute to the sEPSCs, the vast majority require α3*-AChRs because α-Conotoxin-MII, which blocks α3*- but not α7-nAChRs on CG neurons (Nai et al., 2003) reduced sEPSC frequency by 95% (Chen et al., 2001). Exposure to BDNF substantially increased the overall frequency of sEPSCs (Fig. 7), most of which display slow decay kinetics indicative of a major contribution from α3*-nAChRs (Chen et al., 2001). After 16–24 hr BDNF treatment, sEPSC frequency increased approximately threefold (2.69 ± 0.35; N = 46) relative to untreated control neurons from the same five cultures tested in parallel (1.00 ± 0.17; N = 39), with a smaller yet significant increase seen after 4–5 d treatment (Fig. 7A,B). This effect is reminiscent of that seen at other synapses, where BDNF increases EPSC frequency by a presumed presynaptic mechanism (McAllister et al., 1999; Schnider and Poo, 2000; Poo, 2001) (see below). To determine if BDNF also altered sEPSC amplitudes, well separated individual synaptic currents were extracted from selected records, and components mediated by α7- and α3*-nAChRs were identified by their diagnostic fast and slow decay kinetics, as previously described (Chen et al., 2001). The amplitudes of fast, α7-nAChR-mediated sEPSCs identified in this manner increased after 16–24 hr of BDNF treatment (Fig. 7D), shifting by 32% from a median value of –9.6 pA in controls to –12.7 pA in BDNF-treated cultures (N = 4 neurons each; p < 0.0004; Mann–Whitney U and Kolmogorov–Smirnov tests). The effect was selective for α7-nAChR-mediated sEPSCs because, despite increasing in frequency, slow α3*-nAChR-mediated sEPSCs displayed amplitudes that were unchanged by BDNF treatment (Fig. 7E). Recent studies indicate that chronic exposure to BDNF increases the proportion of postsynaptic α7-nAChR clusters on hippocampal neurons (Kawai et al., 2002). Because α3*-nAChR mediated sEPSC amplitudes were unchanged, a similar postsynaptic accumulation of α7-nAChRs may also explain the larger amplitude fast sEPSCs seen here after 16–24 hr exposure to BDNF.
Although significant and α7-nAChR-selective, the changes in fast sEPSC amplitudes after 16–24 hr BDNF treatment were small in comparison to the accompanying threefold increase in the frequency of (primarily) α3*-nAChR-mediated sEPSCs. Studies in other systems suggest that this latter, more dramatic effect is likely to be presynaptic in origin, possibly resulting from changes in intracellular Ca2+ dynamics that alter quantal release (Pozzo-Miller et al., 1999; Tyler et al., 2002). Interestingly, presynaptic α7-nAChRs enhance neurotransmitter release and are known to do so by elevating terminal Ca2+ levels (Gray et al., 1996; Coggan et al., 1997), possibly through Ca2+-induced Ca2+-release (CICR) recently shown to increase EPSC frequency (Sharma and Vijayaraghavan, 2003). Because CG neurons in culture express Ca2+-permeable α7-nAChRs on neurite tips (Pugh and Berg, 1994), and activation of CICR markedly enhances sEPSC frequency (M. Chen and J. Margiotta, unpublished observations), we wondered if upregulation of presynaptic α7-nAChRs might underlie the ability of BDNF to increase sEPSC frequency. This hypothesis predicts that BDNF applied for 10–30 min should not increase sEPSC frequency because brief exposures were insufficient to increase surface α7-nAChR levels or somatic α7-nAChR currents (Figs. 5, 6). In accord with results from other systems (McAllister et al., 1999; Schnider and Poo, 2000; Poo, 2001) however, brief exposure to BDNF induced a significant, K252a-sensitive increase in sEPSC frequency (Fig. 7C, left), thereby demonstrating an expected trk dependence, but arguing against a requirement for rapid α7-nAChR modulation. Because α7-nAChRs at somatic and presynaptic sites could differ in their acute responsiveness to BDNF, we devised a more direct test, blocking α7-nAChRs with α-Bgt and comparing sEPSCs in cultures treated with or without coapplied BDNF. Even with α-Bgt present to block α7-nAChRs, however, BDNF applied for 16–24 hr was still able to increase sEPSC frequency (Fig. 7C, right), with all events now displaying slow decay kinetics indicative of α3*-nAChRs. These results indicate that brief- (10–30 min) and intermediate-duration exposures to BDNF (16–24 hr) can increase sEPSC frequency and do so without a requirement for α7-nAChRs. Nevertheless, α7-nAChRs are strongly implicated in long-term synaptic regulation (Role and Berg, 1996; MacDermott et al., 1999; Liu et al., 2001; Kawai et al., 2002). Thus, because 4–5 d BDNF treatments also increased sEPSC frequency and were required to detect significant changes in α7-nAChRs, we cannot exclude the possibility that chronic neurotrophin exposure sustains the long-term function of neuronal nicotinic synapses in ways that somehow depend on α7-nAChRs.
Discussion
Detection of trkB and BDNF
We have shown that trkB and BDNF-like proteins are present in the chick CG, a model parasympathetic system, where both trks and neurotrophins were presumed irrelevant. No other studies have attempted to detect trkB protein in CG, however, previous Northern and RNase protection analyses failed to detect trkB mRNA (Dechant et al., 1993; Hallbook et al., 1995), possibly because of the lower sensitivity of these assays compared with RT-PCR. Standard criteria for trkB primer (Garner and Large, 1994; Garner et al., 1996) and antiserum specificity (von Bartheld et al., 1996), as well as controls involving primer and primary antiserum omission (this study) support our detection of trkB mRNA in CG, and cell surface trkB protein on CG neurons. In addition, the observations that BDNF elicits trk-dependent signaling leading to CREB activation and trk-dependent changes in α7-nAChRs and synaptic function further indicate that CG neurons express functional trkB. Because the trkB antiserum used recognizes an extracellular epitope (von Bartheld et al., 1996), and PCR amplifications using F/RK primers revealed the presence of isoforms lacking an intracellular kinase domain, however, some trkB immunoreactivity may represent truncated receptor. Whereas the role of kinase-deficient trk isoforms is poorly understood, the notion they are expressed on CG neurons is strengthened by the observation that in 50% of neurons BDNF application failed to induce p-CREB, a process expected to require trkB kinase activity (Finkbeiner et al., 1997; Huang and Reichardt, 2003). In such cases, full-length trkB receptors may still be present but either rendered functionally incompetent or expressed at levels insufficient to activate CREB because truncated trk isoforms have been reported to inhibit both the function and expression of full-length receptors (for review, see Huang and Reichardt, 2003).
BNDF–trkB signaling upregulates α7-nAChRs
BDNF treatment for 4–5 d induced trk-dependent increases in α7 subunit mRNA and surface α7-nAChRs, and enhanced α7-nAChR-mediated whole-cell currents, all without changing levels of α3 subunit mRNA or α3*-nAChR-mediated currents. Similarly, NGF has been shown to selectively increase expression of α7- over α3-nAChR subunit mRNA in sympathetic neuron-like PC12 cells (Takahashi et al., 1999; but see Henderson et al., 1994). Although alterations in receptor turnover rates and mRNA stability may contribute, a straightforward interpretation of our results is that BDNF activation of trkB leads to increased α7-nAChR subunit transcription and protein synthesis, thereby increasing levels of assembled cell-surface receptor. One way BDNF–trkB signaling may influence α7-nAChR subunit transcription is through activation of transcription factors including not only CREB (Finkbeiner et al., 1997), but also AP-1, or NF-κB, which like CREB are reported to be stimulated by BDNF–trkB signaling (Gaiddon et al., 1996; Lipsky et al., 2001). CRE binding sites are present in promoter-containing regions of human and bovine α7-nAChR subunit genes, although in the chicken gene a 1298 bp 5′ segment with a basal promoter at –406 to –230 was previously reported to lack a strong consensus CRE binding site (Matter-Sadzinski et al., 1992; Gault et al., 1998). Within this same 5′ region, however, a new search of two transcription factor databases [Transfac (http://www.gene-regulation.com/pub/databases.html; Heinemeyer et al., 1998) and Matinspector (http://www.genomatix.de/freelogin.htm)] did reveal a potential (–) strand CRE binding site (T-1060GACcTAA-1067) upstream from the basal promoter. Potential binding sites for AP-1 (T-715TcACTCAG-708) and NF-κB(G-176GGGgcTCCC-167) were also predicted in the 5′-flanking and basal promoter regions, respectively. These considerations suggest that BDNF–trkB signaling can regulate the chicken α7-nAChR subunit gene via CRE, AP-1, or NF-κB. Without experimental data, however, it is difficult to judge the significance of these transcription factors as direct regulators. Here, it should be noted that binding sites for Egr-1, Sp1, and Sp3, transcription factors not associated with BDNF–trkB signaling, are thought to regulate the activity of the rat α7-nAChR promoter (Nagavarapu et al., 2001).
BNDF increases activity at nicotinic synapses on CG neurons
BDNF increased sEPSC frequency after acute (10–30 min), intermediate (16–24 hr), or long-term (4–5 d) treatments. The increased sEPSC frequency after acute BDNF exposure depended on trkB activation and resembled that seen at other peripheral and central synapses, where enhanced transmitter release from presynaptic terminals is implicated (McAllister et al., 1999; Schnider and Poo, 2000; Poo, 2001). The basis of the acute synaptic effects seen here and in these other systems is unknown, but seems likely to reflect BDNF actions on Ca2+ (Berninger et al., 1993; Stoop and Poo, 1996; Li et al., 1998) and vesicular dynamics (Pozzo-Miller et al., 1999; Tyler et al., 2002) in presynaptic terminals that enhance neurotransmission reliability. The compelling possibility that Ca2+-permeable, presynaptic α7-nAChRs underlie these effects (Gray et al., 1996; Coggan et al., 1997; Sharma and Vijayaraghavan, 2003) is unlikely, however, because acute BDNF exposure failed to modulate somatic α7-nAChR currents and, more telling, because coincubation with α-Bgt failed to block the ability of BDNF to increase sEPSC frequency. Also unlikely are general effects on membrane excitability as seen for PC12 cells (Rudy et al., 1987; Lesser et al., 1997) because BDNF treatments failed to detectably change the amplitude or voltage sensitivity of somatic Na+ or Ca2+ currents. In addition to increasing overall sEPSC frequency threefold, 16–24 hr BDNF treatments specifically increased the amplitude of α7-nAChR-mediated sEPSCs. Although we cannot exclude increased quantal release at presynaptic nerve terminals that contact only α7-nAChR clusters, this effect seems more likely to be postsynaptic in origin. Unlike currents induced by rapid nicotine microperfusion, which represent nAChR function integrated over the entire soma and report only a nominal increase, sEPSCs are focal events, and an increase in their amplitude would be expected even after adding a few functional receptors in the postsynaptic membrane. The increase in α7-nAChR-mediated sEPSC amplitudes agrees well with the increased postsynaptic α7-nAChR clusters previously observed in hippocampal neuron cultures (Kawai et al., 2002) and with the nominal increase in [125I]-α-Bgt binding seen here after 16–24 hr exposure to BDNF and the significant increase seen after 5 d. The elevated sEPSC amplitude could reflect increased α7-nAChR synthesis, and/or preferential insertion at existing postsynaptic sites, but we cannot presently distinguish between these possibilities.
Long-term synaptic enhancement
Our findings indicate that acute- and intermediate-term BDNF treatments increased synaptic activity without a requirement for α7-nAChRs. Chronic (4–5 d) BDNF treatments continued to enhance synaptic activity, however, and, in parallel, significantly increased α7-nAChR levels and whole-cell currents. α7-nAChRs have been linked to activity-dependent neurite outgrowth and other developmental processes (for review, see Margiotta and Pugh, 2004) that may normally help ensure appropriate synaptogenesis or sustain existing functional synapses once formed (for review, see Role and Berg, 1996; Broide and Leslie, 1999; Jones et al., 1999). In addition, BDNF has been shown to have potent long-term effects on synaptic development and maintenance in other systems (for review, see Poo, 2001). Thus, although further experiments are needed, it remains possible that, in contrast to short- and intermediate-term α7-nAChR-independent effects, the ability of BDNF signaling to sustain long-term synaptic function is related somehow to coincident regulation of α7-nAChRs.
In vivo relevance?
Our results do not directly address the relevance of signals generated by BDNF through trkB for CG neurons in vivo. That targets and target-derived factors influence the survival, growth, and differentiation of input neurons, however, has been recognized for decades (Berg, 1984; Levi-Montalcini, 1987). Specific to this report, previous studies demonstrated that synapses on CG neurons undergo patterned maturation between E8 and E18 and that normal neuron survival and ganglionic transmission require connection to the intraocular muscle targets (Landmesser and Pilar, 1974a,b). More recent experiments indicate that α7-subunit mRNA and α7-nAChR protein and currents all increase during the same developmental period (Corriveau and Berg, 1993; Blumenthal et al., 1999) and that severing peripheral target connections reduces levels of α7-mRNA and protein (Brumwell et al., 2002). Given the importance of target connections in sustaining α7-nAChRs and synapses, the presence of BDNF-like protein in the iris target suggests it may influence synaptic properties of CG neurons during development in vivo. The precise spatial and temporal patterns of BDNF and trkB expression still need to be determined. Our PCR and ELISA results suggest, however, that BDNF is both synthesized within the CG, perhaps by the neurons themselves, and transported to the ganglion from the iris muscle. Such local BDNF expression and retrograde transport from the target are consistent with the arrangement in spinal cord (Koliastsos et al., 1993) and would concentrate BDNF in the iris and CG relative to its levels in eye extract, as was observed here. In addition to BDNF, NGF and NT-3 signaling may also be important in the CG system because recent findings indicate mRNAs for trkA and trkC are expressed in the ganglion, and CG neurons express trkA and trkC immunoreactivity (Dittus et al., 2002). Experiments are in progress to reassess the ability of NGF, NT-3, and BDNF to promote CG neuronal survival, and to identify their respective roles in regulating nAChR expression and nicotinic synaptic differentiation on these parasympathetic neurons.
Footnotes
This work was supported by National Institutes of Health Grants R01-DA15536 (J.F.M.) and R01-NS40644 (M.J.H.). We thank Wei Han for technical assistance, Drs. Frances Lefcort and Louis Reichardt for providing trkB and p75NTR antisera, Dr. Xuan Zheng for advice on real-time PCR, and Drs. Darwin Berg, Leslie Henderson, and Phyllis Pugh for helpful comments on this manuscript.
Correspondence should be addressed to Dr. Joseph F. Margiotta, Medical College of Ohio, Department of Anatomy and Neurobiology, BHS 108, 3035 Arlington Avenue, Toledo, OH 43614-5804. E-mail: jmargiotta{at}mco.edu.
Q. Nai's present address: Department of Biology 0357, University of California, San Diego, 9500 Gilman Drive, La Jolla, CA 92093.
M. Chen's present address: Behavioral Medicine Research Institute, Ohio State University, 2187 Graves Hall, 333 West Tenth Avenue, Columbus, OH 43210.
DOI:10.1523/JNEUROSCI.0055-04.2004
Copyright © 2004 Society for Neuroscience 0270-6474/04/244340-11$15.00/0
↵* Q.N. and M.C. contributed equally to this work.