Enhanced BDNF signalling following chronic hypoxia potentiates catecholamine release from cultured rat adrenal chromaffin cells

Abstract

Environmental stressors, including chronic hypoxia, enhance the ability of adrenomedullary chromaffin cells (AMCs) to secrete catecholamines; however, the underlying molecular mechanisms remain unclear. Here, we investigated the role of brain-derived neurotrophic factor (BDNF) signalling in rat AMCs exposed to chronic hypoxia. In rat adrenal glands, BDNF and its tropomyosin-related kinase B (TrkB) receptor are highly expressed in the cortex and medulla, respectively. Exposure of AMCs to chronic hypoxia (2% O₂; 48 h) in vitro caused a significant increase to TrkB mRNA expression. A similar increase was observed in an immortalized chromaffin cell line (MAH cells); however, it was absent in MAH cells deficient in the transcription factor HIF-2α. A specific TrkB agonist, 7,8-dihydroxyflavone (7,8-DHF), stimulated quantal catecholamine secretion from chronically hypoxic (CHox; 2% O₂) AMCs to a greater extent than normoxic (Nox; 21% O₂) controls. Activation of TrkB by BDNF or 7,8-DHF increased intracellular Ca²⁺ ([Ca²⁺]_i), an effect that was significantly larger in CHox cells. The 7,8-DHF-induced [Ca²⁺]_i rise was sensitive to the tyrosine kinase inhibitor K252a and nickel (2 mm), but not the Ca²⁺ store-depleting agent cyclopiazonic acid. Blockade of T-type calcium channels with TTA-P2 (1 μm) or voltage-gated Na⁺ channels with TTX inhibited BDNF-induced [Ca²⁺]_i increases. BDNF also induced a dose-dependent enhancement of action potential firing in CHox cells. These data demonstrate that during chronic hypoxia, enhancement of BDNF-TrkB signalling increases voltage-dependent Ca²⁺ influx and catecholamine secretion in chromaffin cells, and that T-type Ca²⁺ channels play a key role in the signalling pathway.

Key points

• We investigated the role of the neurotrophin BDNF signalling via the TrkB receptor in rat adrenomedullary chromaffin cells (AMCs) exposed to normoxia (Nox; 21% O₂) and chronic hypoxia (CHox; 2% O₂) in vitro for ∼48 h.

• TrkB receptor expression was upregulated in primary AMCs and in immortalized chromaffin (MAH) cells exposed to CHox; this effect was absent in MAH cells deficient in the transcription factor, hypoxia inducible factor (HIF)-2α.

• Relative to normoxic controls, activation of the TrkB receptor in chronically hypoxic AMCs led to a marked increase in membrane excitability, intracellular [Ca²⁺], and catecholamine secretion.

• The BDNF-induced rise of intracellular [Ca²⁺] in CHox cells was sensitive to the selective T-type Ca²⁺ channel blocker TTA-P2 and tetrodotoxin (TTX), suggesting key roles of low threshold T-type Ca²⁺ and voltage-gated Na⁺ channels in the signalling pathway.

Introduction

The sympathoadrenal system functions to maintain homeostasis over a broad range of environmental stressors via the release of catecholamines (CAT). During acute exposures to low O₂ (hypoxia), activation of the sympathetic nervous system ensures O₂ supply to vital organs, in part by increasing cardiac output and systemic arterial blood pressure (Marshall, 1994). Elevated sympathetic activity also occurs in both healthy adults and patients experiencing chronic hypoxaemia, in association with increased plasma and urinary catecholamines (Calbet, 2003). While adrenomedullary chromaffin cells (AMCs) are thought to contribute to CAT secretion in relation to the degree and duration of the hypoxic stress (Cannon & Hoskins, 1911; Johnson et al. 1983), intrinsic changes to chromaffin cell function may occur independently of enhanced sympathetic efferent activity. For instance, direct exposure of primary adult rat AMCs to chronic hypoxia in vitro is known to cause a hypoxia inducible factor (HIF)-dependent upregulation of T-type calcium channels and enhanced low-threshold CAT secretion, independent of splanchnic nerve activity (Carabelli et al. 2007).

In the present study, we considered the possibility that changes in neurotrophin signalling contribute to the potentiation of adrenal CAT secretion following chronic hypoxia. Members of the neurotrophin family, consisting of nerve growth factor (NGF), brain-derived neurotrophic factor (BDNF), neurotrophin-3 (NT3) and neurotrophin 4/5 (NT4/5), signal selectively via tropomyosin-related kinase receptors, TrkA, TrkB and TrkC, or universally through the p75 neurotrophin receptor (p75NTR) (Skaper, 2012). Importantly, levels of circulating BDNF in humans significantly increase during hypoxic stress, both in early development (Nikolaou et al. 2006) and adulthood (Hubold et al. 2009; Helan et al. 2014). Likewise, in the adrenal gland the BDNF receptor TrkB is expressed across all developmental periods, appearing in both neonatal (Schober et al. 1999) and adult AMCs (Kondo et al. 2010). The interaction of BDNF with the TrkB receptor is known to regulate synaptic transmission and ion channel activity in many different neuronal tissues (Poo, 2001).

In order to test for a potential role of neurotrophins during chronic hypoxia, we compared BDNF-TrkB signalling in young (∼10-day-old) and more mature (3- to 4-week-old) rat AMCs cultured for 2 days under normal (21% O₂) and low oxygen (2% O₂) conditions. Interestingly, we found that not only was TrkB expression upregulated in low oxygen, activation of this receptor in chronically hypoxic AMCs from both age groups led to a significant potentiation of voltage-gated Ca²⁺ entry and CAT secretion. We also tested whether the effects of hypoxia on TrkB expression, as well as the BDNF-induced responses, in AMCs involved the full-length (TrkBFL) or truncated (TrkBT1) isoform of the receptor (Middlemas et al. 1991). Finally, because the transcription factor HIF-2α is known to play a predominant role in the hypoxic response of cells of the sympathoadrenal lineage (Tian et al. 1998; Bishop et al. 2008; Brown et al. 2009; Richter et al. 2013), we compared TrkB expression in a control versus HIF-2α-deficient immortalized chromaffin cell line (Brown et al. 2009) derived from the fetal rat adrenal gland. The combined data revealed a novel role for BDNF–TrkB signalling in chromaffin cells during hypoxic stress and point to TrkB as a target for HIF-2α.

Methods

Ethical approval

All procedures for animal handling and tissue isolation were carried out according to the guidelines of the Central Animal Facilities at McMaster University and the Canadian Council on Animal Care (CCAC), and comply with the policies and regulations of The Journal of Physiology (Drummond, 2009).

Primary chromaffin cell cultures

Methods for preparing cultures of dissociated rat chromaffin cells were similar to those previously used in this laboratory (Thompson & Nurse, 1998). Wistar rats were provided by Charles River (Quebec, Canada) and housed in the Central Animal Facility at McMaster University. Briefly, juvenile rat pups, 10–12 days old, were quickly rendered unconscious by a swift blow to the head and killed immediately by decapitation. Older rats (3–4 weeks old) were given an overdose of halothane via inhalation and then underwent cervical dislocation. The adrenal glands from either age group were then bilaterally dissected from the animals and placed in L-15 plating medium (Gibco, Grand Island, NY, USA), and much of the surrounding adrenal cortex was removed and discarded. The remaining medullary-enriched tissue was incubated in 0.1% trypsin (Gibco) and 0.1% collagenase (Gibco or Sigma-Aldrich, Oakville, Canada) at 37°C for 50 min, followed by mechanical dissociation. The dissociated cells were plated onto modified 35 mm culture dishes that were coated with a thin layer of Matrigel (Collaborative Research, Bedford, MA, USA). The cells were maintained in F-12 nutrient medium (Gibco) that was supplemented with 5% fetal bovine serum, 1% penicillin–streptomycin, 1% glutamine, 0.3% glucose, 5 μm dexamethasone and 3 μg ml⁻¹ insulin. Cultures were incubated in a humidified atmosphere of 95% air–5% CO₂ (normoxia) or 2% O₂–5% CO₂ (hypoxia) for 48 h at 37°C. For each experiment, dissociated adrenomedullary chromaffin cells (AMCs) from one litter of 10–12 pups were divided into two equal fractions, one for normoxia and the other for hypoxia. The cells from each fraction were then plated at roughly similar densities into three to four culture dishes. For data analysis, the indicated (n) values in the text refer to the number of litters used for each experimental protocol.

MAH cell cultures

The v-myc immortalized adrenomedullary chromaffin-derived cells (MAH cells) were grown in L15/CO₂ medium (Gibco), containing 0.6% glucose, 1% penicillin–streptomycin, 10% fetal bovine serum and 5 μm dexamethasone, as previously described (Fearon et al. 2002; Brown et al. 2009). Cells were plated on 35 mm culture dishes that were coated with poly-d-lysine and laminin, fed every 2–3 days, and passaged every 3–4 days. The stable HIF-2α knockdown (shHIF-2α) and scrambled negative control (scControl) cell lines were generated previously in this laboratory using oligonucleotides containing a short hairpin RNAi sequence; shHIF-2α MAH cells consistently show >90% knockdown of HIF-2α protein in Western blot analyses (Brown & Nurse, 2008; Brown et al. 2009; Salman et al. 2012). Similar to primary AMCs, all MAH cell lines were incubated in either a humidified normoxic or hypoxic atmosphere for 48 h at 37°C. By 48 h, the MAH cells were 70–80% confluent and three to four culture dishes were pooled for each experimental replicate (n) and processed for RNA extraction.

Immunofluorescence

Rat pups (P10–12) were given an intraperitoneal injection of sodium pentobarbital (5.5 mg ml⁻¹; 0.4 ml) and perfused with 4% paraformaldehyde. The adrenal glands were dissected out and post-fixed in the paraformaldehyde overnight. Adrenal glands were then submerged in 30% sucrose in 1× phosphate buffered saline (PBS: 150 mm NaCl, 15 mm NaH₂PO₄, pH 7.4) at 4°C for 24 h, and then flash frozen with liquid nitrogen in Histo prep medium (Fisher Scientific, Pittsburgh, PA, USA) and stored at −80°C. Transverse cryosections (18 μm) of the adrenal glands were washed in PBS (3 times, 3 min) and incubated overnight in 1% bovine serum albumin (BSA), PBS, 0.5% Triton X-100 and primary antibodies including tyrosine hydroxylase (TH; 1:4500, host mouse; cat. no. MAB318; Chemicon, Temecula, CA, USA), brain-derived neurotrophic factor (BDNF; 500 μg ml⁻¹; host rabbit; cat. no. 710306; Novex, Carlsbad, CA, USA), and tropomyosin-related tyrosine kinase receptor B (TrkB; 1:100; host rabbit; cat. no. ANT-019; Alomone Labs, Jerusalem, Israel). Sections were then washed in PBS and incubated for 1 h in secondary antibodies conjugated to either Alexa 488 or 594 (both used at 1:400; Molecular Probes, Eugene, OR, USA). Following three 3 min washes in PBS, Vectasheild (Vector Laboratories, Burlington, Canada) was applied to the slides for coverslipping. Primary antibodies were excluded from the solutions as negative controls for both BDNF and TrkB antibodies. The pre-absorption control for TrkB expression was performed by incubating the slides in primary antibody in the presence of TrkB antigen (ratio 1:3) for 48 h prior to secondary antibody exposure.

For immunocytochemical detection of TrkB on primary chromaffin cells, cultures were rinsed in pre-warmed PBS and fixed in 4% paraformaldehyde for 1 h at room temperature. Cells were then washed 3×3 min and then incubated in 1% BSA, PBS, 0.5% Triton X-100 and primary antibodies for TH (1:4500; host mouse; Chemicon) and TrkB (1:100; host rabbit; Alomone Labs) overnight at 4°C. The following day the cells were washed again (3 times for 10 min) in PBS and incubated for 1 h in secondary antibodies conjugated to Alexa 488 and 584 (1:400; Molecular Probes) at room temperature. The glass coverslips containing the cells were then removed from the culture dishes and mounted on slides using the mounting medium Vectashield as described above.

Specimens were examined using the Leica TCS SP5 II confocal system, which was equipped with argon (458476 488, 515 nm) and helium–neon (543, 594, 633 nm) lasers. Images were observed with 20×/0.7 and 63×/1.4 NA oil objectives, captured with a digital camera and processed with LAS AF (version 2.1.2; Leica). All settings for image capture were kept consistent across all sections for each fluorochrome. Images for figures were assembled in Adobe Photoshop CS3 Extended (version 10.0).

Image analysis of TrkB⁺/TH⁺ cells was done following image processing in ImageJ (version 1.46). Images were passed through a threshold set at a value that only highlighted cells with an immunoreactivity level that was significantly (∼10 fold) higher than background levels. This resulted in a conservative estimate of cells that were co-labelled with TrkB and TH antibodies in each group. The highlighted cells were counted and expressed as a proportion of all cells present following exposure to either normoxia (n = 5 groups; 309 cells) or hypoxia (n = 4; 297 cells).

Quantitative PCR

RNA was extracted from the adrenal glands of either rat pups or cell cultures (primary or MAH cells). Tissue from a litter of ∼10-day-old rat pups was pooled for each experimental replicate (n) tested. In cell culture experiments with different treatment groups, each experimental sample consisted of pooled cells taken from one litter of pups (primary cells) or three to four culture dishes of MAH cells at 80% confluence. The tissues or cells were then manually homogenized with a 1 ml glass homogenizer (3 repetitions of 10/sample) in RLT buffer and β-mercaptoethanol, and RNA was extracted according to the procedures outlined in the RNeasy Mini Kit (Qiagen, Mississauga, Canada). RNA quantity and quality were determined (Nanodrop 1000; ThermoScientific, Rockford, IL, USA), normalized to control samples and stored at −80°C. Following the generation of cDNA from 10 μg of RNA of each sample (Superscript III cDNA synthesis kit; Invitrogen), the cDNA was then normalized to control samples and stored at −20°C. The qPCR reaction was made up of 5 μg of cDNA, 0.2 μm forward and reverse primers (listed below) and the DyNAmo SYBR Green qPCR mix (Thermoscientific). The following primers were supplied by MOBIXlab (McMaster University, Hamilton, Ontario, Canada) and designed using Primer 3 to amplify ntrk2 (tropomysosin-related kinase receptor type 2 full length; TrkBFL), forward 5′-ATC TTC ACC CAC CTC AAA CC-3′, reverse 5′-GAA ACC ATT CTC CCC GAA AC-3′; truncated ntrk2 (tropomyosin-related kinase receptor, truncated isoform type 1; TrkBT1), forward 5′-GGG GCT GTG CTG CTT GGT-3′, reverse 5′-GCT GCG GAC ATC TTT GGA GA-3’; ngfr (p75^NTR), forward 5′-CAG TAC AGT GGC GGA TAT GG-3′, reverse 5′-CAG CCA AGA TGG AGC AAT AG-3′; bdnf (brain-derived neurotrophic factor), forward 5′-TGA AAG AAG CAA ACG TCC AC-3′, reverse 5′-GTC GTC AGA CCT CTC GAA CC-3′; and housekeeping genes 18S (18S ribosomal RNA), forward 5′- CCA GTA AGT GCG GGT CAT AAG C-3′, reverse 5′-CCT CAC TAA ACC ATC CAA TCG G-3′; and β-actin, forward 5′-CCT GRA TGC CTC TGG TCG TA-3′, reverse 5′-CCA CT CTT GCT CGA ACT CT-3′. Primer sequences for ntrk2, truncated ntrk2 and bdnf matched those previously published (Kondo et al. 2010, 2012). Amplification of products was carried out using two-step amplification with the Stratagene Mx3000P™ QPCR System (SABiosciences, Mississauga, Canada) and analysis was done using Mx3000P™ software version 2.0 (SABiosciences) and Graphpad Prism (Graphpad Software, Inc., La Jolla, CA, USA). The C_T value for each gene of interest was normalized to the C_T values of two separate housekeeping genes (18S and β-actin) from the sample master mix and measured on the same reading plate. Each plate contained wells for the negative controls including one with no template and one with no SYBR mix, as well as a positive control (brain tissue). Each experimental replicate (n) was repeated in three technical replicates and the results of the technical replicates were averaged for each sample.

Carbon fibre amperometry

The vesicular release of catecholamines from chromaffin cells was detected with carbon fibre amperometry as previously described (Livermore et al. 2011). Cultured P10–12 rat chromaffin cells, exposed to either normoxic or hypoxic conditions for 48 h, were perfused with a bicarbonate-buffered solution alone or in the presence of either TrkB agonist 7,8-dihydroxyflavone (7,8-DHF; 500 nm; R&D Systems, Minneapolis, MN, USA) for 2 min or 30 mm KCl for 30 s. A polarized (+800 mV) carbon fibre electrode (Dagan Corp., Minneapolis, MN, USA) was placed gently onto the cell surface and then retracted slightly for recording. The electrode was attached to a CV203BU headstage and the signal was amplified with Axopatch 200B (Molecular Devices, Sunnyvale, CA, USA). The analog signal was recorded at 10 kHz using Digidata 1322A software, and analysed with Clampex version 9.2. The baseline noise level measured approximately 2 pA. Events that measured less than 2.6 standard deviations above the baseline noise and less than 0.5 milliseconds of duration were discounted for analysis.

The integrated area of secretory events, representative of the number of oxidizable catecholamines (Q), was summated to calculate secretion rate within a fixed time of drug application (fC min⁻¹). The quantal frequency (events min⁻¹) and quantal charge (integrated area of each event, fC) of secretory events in response to 7,8-DHF were also compared in normoxic and hypoxic cells. Only cell recordings with events that were at least 2× standard deviation above baseline during exposure to 7,8-DHF and high KCl (positive control) were used for further analysis. In total, 22 culture dishes (1–2 cells per dish sampled; 1–2 dishes used per litter) were recorded from in each group. Owing to the disproportional number of cells responding to 7,8-DHF between the groups, recordings from four dishes (isolated from 4 litters, 1 dish per litter) were used from the normoxic group and recordings from 13 dishes (isolated from 9 litters, ∼1–2 dishes per litter) were used from the hypoxic group for analysis. Recordings from multiple cells in individual dishes were averaged for each dish.

Calcium imaging

Ratiometric Ca²⁺ imaging techniques were used to measure changes to intracellular free Ca²⁺ concentrations ([Ca²⁺]_i) in cultured 10- to 12-day-old or 3- to 4-week-old chromaffin cells. Following a 48 h incubation in either normoxic or hypoxic conditions, chromaffin cell cultures were pre-loaded with the Ca²⁺ indicator, fura-2 AM (2.5 μm; Molecular Probes), for 30 min at 37°C. Cultures were then continuously perfused with a standard bicarbonate-buffered solution (24 mm NaHCO₃, 115 mm NaCl, 10 mm glucose, 12 mm sucrose, 5 mm KCl, 2 mm CaCl₂, 1 mm MgCl₂) in the presence or absence of the following agents: recombinant human brain-derived neurotrophic factor (BDNF; 100 ng ml⁻¹ or 3.7 nm; Gibco, Life Technologies, Burlington, ON, Canada), 7,8-DHF (500 nm; R&D Systems, Minneapolis, MN, US), K252a (100 nm; Tocris Bioscience, Minneapolis, MN, USA), cyclopiazonic acid (CPA; 500 nm; Tocris Bioscience), nickel (II) chloride hexahydrate (2 mm; Sigma Aldrich, Saint Louis, MO, USA), TTA-P2 (1μm; Alomone Labs), tetrodotoxin (TTX; 500 nm; Abcam Inc., Toronto, ON, Canada) or 30 mm KCl. The extracellular solution was kept at 37°C and the pH was held constant at ∼7.4 by bubbling with a 5% CO₂ gas mixture. In each protocol, an experimental sample (n) consisted of averaged responses from one to two recordings made from different cells within a single dish (10–20 cells sampled per dish). Recordings were obtained from at least five different dishes prepared from a minimum of three to four individual litters (1–2 dishes per litter).

Measurements of [Ca²⁺]_i were obtained as previously described in this laboratory (Piskuric & Nurse, 2012). The imaging system consisted of a Nikon Eclipse TE2000-U inverted microscope with a Nikon S-Fluor 40× oil-immersion objective (Nikon, Mississauga, Canada), a Lambda DG-4 ultra high-speed wavelength changer (Sutter Instrument Co., Novato, CA, USA), and a Hamamatsu OCRCA-ET digital CCD camera (Hamamatsu, Sewickley, PA, USA). Simultaneous images were acquired at 340 nm and 380 nm excitations (510 nm emission) every 2 s, with an exposure time of 100 ms. Ratiometric measurements were collected using Simple PCI software version 5.3 and used to calculate the [Ca²⁺]_i according to the Grynkiewicz equation (Grynkiewicz et al. 1985). The intracellular calcium transients were expressed as either the integrated [Ca²⁺]_i response over time (cumulative area above baseline) during single drug treatments, or the average [Ca²⁺]_i concentration (average response above baseline over a specified time period) when agonists and antagonists were given within the same experiment.

Electrophysiology

Electrophysiological recordings from normoxic and chronically hypoxic chromaffin cells were obtained using the perforated patch-clamp technique as previously described (Thompson & Nurse, 1998; Buttigieg et al. 2008; Salman et al. 2012). Membrane potential measurements were made in current clamp mode. The pipette solution contained potassium gluconate (115 mm), KCl (25 mm), NaCl (5 mm), CaCl₂ (1 mm), Hepes (10 mm, pH 7.2), and nystatin (300–450 μg ml⁻¹). The extracellular solution was maintained at 37°C and consisted of 115 mm NaCl, 5 mm KCl, 2 mm CaCl₂, 2 mm MgCl₂, 10 mm glucose, and 24 mm NaHCO₃; pH was maintained at 7.4 by bubbling with 5% CO₂. In most experiments, either BDNF (100 ng ml⁻¹) or 7,8-DHF (500 nm) was applied to the cells via a rapid perfusion system and changes in membrane potential were recorded (Buttigieg et al. 2008). For a few experiments, a range of BDNF concentrations (20, 50 and 100 ng ml⁻¹) was applied to the same cell. Only cells with a resting potential more negative than −45 mV were selected for application of TrkB agonists.

Statistics

Statistical analyses were performed using GraphPad Prism (version 5) and data compared using either the Mann–Whitney test for all non-parametric pairwise comparisons or one-way ANOVA for multiple comparisons with Tukey’s post hoc multiple comparison test. The tests used and P values are specified for each figure presented. P < 0.05 was considered to be statistically significant.

Results

The TrkB receptor and its ligand BDNF are highly expressed within the juvenile adrenal gland

The adrenal gland is separated into two distinct regions, the outer cortex and the inner medulla. The outer cortical region is segregated into several zones: zona glomerulosa (outermost layer), zona reticularis (innermost layer) and zona fasciculata (between the glomerulosa and reticularis) as designated in Fig.1A. Immunohistochemical and quantitative PCR techniques revealed detectable TrkB expression in the cortex and medulla; however, significantly higher levels of expression were detectable in the medulla relative to the cortex (Fig.1A and C; n = 3; P = 0.039). Examination of co-expression of TrkB and tyrosine hydroxylase (TH), a catecholaminergic marker that labels chromaffin cells, revealed that although staining intensity for TrkB appeared stronger in subsets of cells within the medulla (Fig.1Ab), the vast majority of TH⁺ chromaffin cells were immunopositive for TrkB (Fig.1Aa and Ab). Specificity of the TrkB antibody was confirmed since pre-incubation of the primary antibody with a specific TrkB receptor blocking peptide abolished all positive immunolabelling (Fig.1B, Ba and Bb).

Figure 1.

Expression of the tropomyosin-related kinase receptor, TrkB, in the juvenile rat adrenal gland

A and C, expression of the TrkB receptor protein (A) and full-length TrkB (TrkBFL) mRNA (C) is highly prevalent in the chromaffin cells within the adrenal medulla, compared to the relatively low level expression within the cortical zones of the adrenal gland (n = 3 litters/group; P = 0.039). Aa and Ab, magnification of the inset (shown in A) shows co-localization of TrkB with TH⁺ chromaffin cells, though immunoreactivity was variable across the section. B, pre-absorption with a TrkB blocking peptide prevented positive TrkB immunostaining of chromaffin cells (Ba and Bb). D, expression levels of truncated TrkB (TrkBT1) in the adrenal medulla are not significantly different from that in the cortex (n = 3 litters/group; P = 0.200). Each n represents pooled tissues isolated from individual rat litters. TH, tyrosine hydroxylase; BV, blood vessels. Asterisk denotes significant differences between groups (Mann–Whitney analysis). Error bars: ±SEM; scale bar: 50 μm.

Notably, the TrkB antibody does not distinguish between the full-length and the truncated form of the receptor. Quantitative analysis of the mRNA expression revealed a differential distribution in the expression pattern between the two isoforms. While the mRNA expression of full-length TrkB receptor (TrkBFL) was significantly greater in the medullary cells compared to the cortical cells (Fig.1C), there was no significant difference in expression of truncated TrkB (TrkT1) mRNA (Fig.1D; n = 3; P = 0.200).

In striking contrast to the expression pattern of the TrkB receptor, its ligand, BDNF, was more highly expressed within the cortical layers of the adrenal gland compared to the medulla, as demonstrated by immunohistochemistry (Fig.2A), and quantitative PCR (Fig.2D; n = 3; P = 0.001). Expression of BDNF within the medulla did not co-localize with TH-positive chromaffin cells and was localized mainly to blood vessels in this region (Fig.2B, Ba and Bb). In control experiments, BDNF immunostaining was absent in both the cortex and medulla (shown separated by the dashed line) of the adrenal gland when the primary antibody was omitted (Fig.2C and Ca). The reciprocal expression pattern of BDNF and TrkB in the cortex and microvasculature versus medullary TH⁺ cells raises the possibility that BDNF signalling in the adrenal gland occurs in a paracrine fashion.

Figure 2.

Expression of brain-derived neurotrophic factor in the juvenile rat adrenal gland

A and B, BDNF immunoreactivity was detectable in the cortex (A) and in association with blood vessels of the medulla (B), where it did not co-localize with TH⁺ chromaffin cells (Ba and Bb). In C, BDNF immunoreactivity was absent in both the cortex and medulla (separated by dashed line; medullary cells are TH⁺ in Ca) when the primary antibody was omitted (n = 3). D, mRNA levels of BDNF were significantly higher in the cortex compared to medulla (n = 3 litters/group; P = 0.001). Each n represents pooled tissues isolated from individual rat litters. BDNF, brain-derived neurotrophic factor; TH, tyrosine hydroxylase. Asterisk denotes significant differences between groups (Mann–Whitney analysis). Error bars: ±SEM; scale bar: 50 μm.

TrkB expression in chromaffin cells is increased following exposure to chronic hypoxia and is regulated by HIF-2α

To test whether the expression of the TrkB receptor is regulated by chronic hypoxia, we first exposed primary cultures enriched in chromaffin cells from 10- to 12-day-old rats to either a normoxic (20% O₂) or a hypoxic (2% O₂) environment in vitro for ∼48 h. Normoxic (Nox) and chronically hypoxic (CHox) chromaffin cells demonstrated similar morphology and immunostaining pattern for tyrosine hydroxylase (TH; red) (Fig.3A and B). However, immunofluorescence staining for TrkB revealed a staining pattern that appeared more intense in CHox compared to Nox chromaffin cells (Fig.3A and B; green label). The number of cells that co-expressed the TrkB receptor and TH was significantly greater in cultures exposed to CHox (n = 4 cultures, 297 cells analysed) than those in Nox (n = 5 cultures, 309 cells analysed) (Fig.3C, P = 0.016).

Figure 3.

Expression of TrkB in cultured adrenal chromaffin cells following exposure to normoxia or chronic hypoxia

A and B, double-label immunofluorescence staining for TH (red) and TrkB (green) in chromaffin cell cultures grown under normoxia (20% O₂) and chronic hypoxia (2% O₂) for 48 h. C, the percentage of TH⁺ chromaffin cells that co-labelled with TrkB was signficantly greater following hypoxia exposure (n = 4 cultures, 297 cells) compared to those in normoxia (n = 5 cultures, 309 cells; P = 0.016). D, TrkB mRNA expression was significantly upregulated in chromaffin cell cultures exposed to chronic hypoxia as compared to those in normoxia (n = 4 litters/group; P = 0.028). Each n represents pooled cells obtained from individual rat litters. Nox, normoxia; CHox, chronic hypoxia; BDNF, brain-derived neurotrophic factor; TH, tyrosine hydroxylase. Asterisk denotes significant differences between groups (Mann–Whitney analysis). Error bars: ±SEM; scale bar: 25 μm.

Quantitative examination of TrkB mRNA expression levels also revealed a significantly higher expression of TrkB mRNA in CHox chromaffin cells compared to Nox cells (Fig.3D; n = 4/group; P = 0.028). The expression levels of other BDNF receptor mRNAs in chromaffin cell cultures, including the short isoform TrkBT1 and the pan neurotrophin receptor (p75^NTR) (Suter-Crazzolara et al. 1996; Kondo et al. 2010), were also tested. In contrast to full-length TrkB, expression levels of TrkBT1 (n = 4/group; P = 0.342) and p75^NTR (n = 4/group; P = 0.686) were unchanged following exposure to chronic hypoxia. Moreover, expression of BDNF mRNA was also not significantly different in Nox versus CHox chromaffin cell cultures (n = 4/group; P = 0.146) (data not shown).

Given the significant increase in full-length TrkB expression following chronic hypoxia, we tested whether this receptor was regulated by the transcription factor hypoxia inducible factor 2α (HIF-2α). Among the HIF family members, HIF-2α is expressed in adrenal chromaffin cells and is known to be a key regulator of the hypoxic response in sympathoadrenal cells (Tian et al. 1998; Bishop et al. 2008; Richter et al. 2013). To facilitate these studies, we used the immortalized v-myc, adrenal-derived, HNK1⁺ (MAH) cell line, derived from fetal chromaffin sympathoadrenal precursors (Birren & Anderson, 1990). Previous studies in this laboratory demonstrated that HIF-2α expression is greatly enhanced in MAH cells after exposure to chronic hypoxia (Brown & Nurse, 2008; Brown et al. 2009). To test whether the upregulation of TrkB expression during CHox is HIF-2α dependent, we compared TrkB expression in cultures of control MAH cells (control; Fig.4A), MAH cells containing a short hairpin RNAi sequence resulting in >90% knockdown of HIF-2α (shHIF-2α; Fig.4B), and MAH cells containing a scrambled negative control sequence (scControl) (Brown & Nurse, 2008). As illustrated in Fig.4C, in control (white bars; n = 4/group) and scControl MAH cells (grey bars; n = 4/group), there was a significant upregulation of TrkB mRNA expression following 48 h of hypoxia exposure (2% O₂, Fig.4C; P = 0.028 and P = 0.015, respectively). In contrast, hypoxia failed to upregulate TrkB expression in HIF-2α-deficient (shHIF-2α) MAH cells (Fig.4C; black bars, n = 3/group; P = 0.983), causing this group to have significantly lower TrkB expression compared to the CHox scControl (P = 0.019). These findings indicate that HIF-2α is required for the upregulation of the TrkB receptor in chromaffin cells during chronic hypoxia.

Figure 4.

Expression of TrkB is regulated by hypoxia inducible factor 2α, HIF-2α, in immortalized chromaffin cells

A and B, phase contrast micrographs of a control v-myc immortalized (MAH) chromaffin cell line (A), and a HIF-2α-deficient (shHIF-2α) MAH cell line (B). C, following 48 h exposure to hypoxia TrkB mRNA expression was significantly upregulated in control cells (white bars; n = 4/group, 16 culture dishes; P = 0.028), and those containing a scrambled silencing RNA sequence (grey bars; n = 4/group, 16 culture dishes; P = 0.015). Knockdown of HIF-2α expression prevented upregulation of TrkB in shHIF-2α MAH cells (black bars; n = 3/group, 12 culture dishes; P = 0.983), which was significantly lower than in scControl CHox cells (P = 0.019). Each n represents cells isolated from individual passage experiments. Nox, normoxia; CHox, chronic hypoxia; MAH, immortalized adrenomedullary chromaffin cells; shHIF-2α, short hairpin HIF-2α knockdown MAH cells. Asterisks denote significant differences between normoxic and hypoxic groups; dagger denotes significant differences within CHox groups (one-way ANOVA). Error bars: ±SEM; scale bar: 10 μm.

Activation of TrkB receptors augments catecholamine secretion in chromaffin cells exposed to chronic hypoxia

In the central nervous system, neurotrophin stimulation of TrkB receptors is known to modulate neurotransmitter release (Mattson, 2008). Given that catecholamine (CAT) secretion is the principal physiological response of adrenal chromaffin cells during sympathetic activation, we wondered whether the enhanced TrkB receptor expression observed in chromaffin cells after chronic hypoxia (CHox; 2% O₂, 48 h) was correlated with greater agonist-induced CAT secretion. Using carbon fibre amperometry, we found that the specific TrkB receptor agonist 7,8-dihydroxyflavone (7,8-DHF; 500 nm) stimulated quantal CAT secretion in subsets of primary chromaffin cells grown in normoxia (Nox; 21% O_2, 48 h) and chronic hypoxia (Fig. 5A and B). Some cells were considered ‘non-responsive’ to 7,8-DHF in cases where events did not exceed 2 standard deviations above baseline (Fig.5C). The fact that these ‘non-responsive’ cells were still able to evoke robust quantal CAT secretion in response to the depolarizing stimulus of high K⁺ (30 mm) suggested that their secretory machinery was intact. In concert with the increased TrkB expression seen in cells exposed to chronic hypoxia, there was a significantly larger proportion of CHox cells (∼60%) that evoked detectable CAT secretion during application of 7,8-DHF compared to Nox cells (∼20%) (n = 22/group; P = 0.008; Fig.5D). Moreover, the 7,8-DHF-induced cumulative secretion rate (fC min⁻¹) in responsive cells was also significantly enhanced (∼6×) in CHox (n = 9) compared to Nox (n = 4) cells (Fig.5E; P = 0.005). This increase in cumulative secretion rate was attributable mainly to an increase in quantal frequency (Fig.5F; P = 0.027), as there were no detectable differences in mean quantal charge (Q; Fig.5G; P = 0.444). These data suggest that the increase in TrkB expression observed in CHox chromaffin cells was associated with a potentiation of CAT secretion during exposure to TrkB agonists.

Figure 5.

Amperometric detection of catecholamine release stimulated by TrkB receptor activation

A–C, 7,8-DHF application (500 nm; 2 min) to either normoxic (21% O₂; 48 h; n = 22, isolated from 9 litters) or hypoxic (2% O₂; 48 h; n = 22, 9 litters) chromaffin cells resulted in the amperometric detection of catecholamine release in some (A and B) but not all (C) cells; note that all cells, including those that were non-responsive to 7,8-DHF (C), did respond to the depolarizing stimulus, high K⁺. D, the proportion of cells that responded to 7,8-DHF was significantly higher in the hypoxic compared to normoxic group (P = 0.008). E, in responsive cells, the secretion rate was significantly higher in the hypoxic group (n = 9, 13 responsive cells) compared to the normoxic (n = 6, 6 responsive cells; P = 0.005) group. F and G, the difference in the overall secretion rate was due to a significant difference in frequency of quantal events (P = 0.027), but not quantal charge (P = 0.444; also see expanded scale below A and B). Each n represents number of culture dishes sampled. Nox, normoxia; CHox, chronic hypoxia; High K⁺, 30 mm KCl. Asterisks denote significant differences between groups (Mann–Whitney analysis). Error bars: ±SEM.

Enhancement of BDNF-induced rise of intracellular [Ca²⁺] via TrkB receptors in chromaffin cells exposed to chronic hypoxia

BDNF activation of the TrkB receptor is known to play a key role in regulating synaptic activity in central neurons via increases in intracellular Ca²⁺ (Levine et al. 1995; Baldelli et al. 2002; He et al. 2005). Therefore, using ratiometric Fura-2 Ca²⁺ imaging, we tested whether BDNF-TrkB signalling in chromaffin cells was associated with increases in intracellular Ca²⁺ ([Ca²⁺]_i) transients and, if so, whether the magnitude of these Ca²⁺ responses was elevated after chronic hypoxia. For these studies, we cultured chromaffin cells obtained from both young (∼10-day-old) rats when innervation of the adrenal gland was still developing, and more mature (3- to 4-week-old) rats when innervation was complete. As exemplified in Fig.6A for both age groups, BDNF (100 ng ml⁻¹) induced a rise in [Ca²⁺]_i in subpopulations of Nox and CHox chromaffin cells, though the responses in CHox cells were significantly larger. Data from cultures isolated from 10- to 12-day-old (filled bars) and 3- to 4-week-old (open bars) rats grown under the two conditions are summarized in Fig.6B. For the cells obtained from 10- to 12-day-old pups, the mean peak BDNF-induced rise in [Ca²⁺]_i (Peak Δ[Ca²⁺]_i) was ∼2× higher in CHox cells (left panel; n = 5; P = 0.028) than Nox cells (n = 5); however, mean basal [Ca²⁺]_i was not significantly different between Nox (127.3 ± 10.4 nm) and CHox (96.2 ± 17.2 nm) cells (P = 0.151). Also, the mean Δ[Ca²⁺]_i response to the depolarizing stimulus of high K⁺ (30 mm) was not significantly different between the Nox (852.2 ± 101.1 nm) and CHox (891.0 ± 75.72 nm) cells, suggesting similar high-threshold voltage-gated Ca²⁺ channel activity in the two conditions (n = 5/group; P = 0.699). The BDNF-induced peak Δ[Ca²⁺]_i was also greater in CHox cells (n = 6) compared to Nox cells (n = 5; P = 0.017) from 3- to 4-week-old rats. In addition to the larger peak [Ca²⁺]_i, the duration of the Ca²⁺ response to the BDNF stimulus was significantly longer (∼2×) in younger and older CHox cells, as summarized in Fig.6B (middle; P = 0.016 and P = 0.026, respectively). Given the higher mean peak and longer-lasting [Ca²⁺]_i response in CHox cells, the integrated area under the Ca²⁺ transient curves is likely to provide a better estimate of the overall effect of BDNF on intracellular Ca²⁺ signalling (Fig.6B, right panel). The integrated Δ[Ca²⁺]_i response in CHox cells from 10-day-old animals was ∼4× larger than that in Nox cells (P = 0.008); and in cells obtained from 3- to 4-week-old animals, the integrated response was ∼2× larger in CHox compared to Nox cells (n = 6/group; P = 0.026).

Figure 6.

BDNF-induced intracellular calcium responses in normoxic versus hypoxic adrenal chromaffin cells

A (left) and B (black bars), BDNF (100 ng ml⁻¹; 1 min) caused a significantly greater increase in peak intracellular [Ca²⁺]_i in chronically hypoxic ∼10-day-old chromaffin cells compared to normoxic controls (B, left panel; n = 5/group, 47 hypoxic cells and 44 normoxic cells; P = 0.028). The mean duration of the Ca²⁺ response (B, middle panel; P = 0.016), as well as the integrated Ca²⁺ response over time (B; right panel; P = 0.008) were also significantly greater for the hypoxic group. A (right) and B (white bars), similarly, in 3- to 4-week-old chromaffin cells, peak Δ[Ca²⁺]_i (left panel) was significantly greater in CHox cells (n = 6/group; 46 cells) than Nox cells (n = 5/group; 27 cells; P = 0.017). The same was true for the mean duration of the Ca²⁺ response (middle panel; P = 0.026) and the integrated Ca²⁺ response (right panel; 0.027). Each n represents number of culture dishes sampled. Nox, normoxia; CHox, chronic hypoxia; BDNF, brain-derived neurotrophic factor; High K⁺, 30mm KCl. Asterisks denote significant differences between groups (Mann–Whitney analysis). Error bars: ±SEM.

Given that chromaffin cells not only express full-length TrkB, but also its short isoform TrkBT1 and the pan neurotrophin receptor (p75^NTR) (Ohira & Hayashi, 2009), we next determined which receptor subtype mediated the effects of BDNF on Ca²⁺ signalling. Consistent with a role for TrkB activation, application of the selective TrkB agonist 7,8-dihydroxyflavone (7,8-DHF; 500 nm), which does not interact with the extracellular or intracellular domains of other Trk receptors or p75^NTR (Jang et al. 2010), resulted in increases to intracellular Ca²⁺ in both Nox and CHox chromaffin cells (Fig.7A). Similar to BDNF, both the magnitude of the 7,8-DHF-induced peak Ca²⁺ responses (Fig.7B; P = 0.013) and the frequency of responders (Fig.7C; P = 0.014) were significantly larger in CHox (n = 14) compared to Nox (n = 7) cells. To distinguish between the full-length TrkB receptor and its short isoform lacking the tyrosine kinase site, we applied 7,8-DHF in conjunction with a tyrosine kinase inhibitor, K252a. In the presence of K252a (100 nm), the 7,8-DHF-induced increases in [Ca²⁺]_i were significantly reduced in CHox cells (n = 14; P = 0.031) and responses to the depolarizing stimulus of high K⁺ appeared normal after wash-out of the drugs (Fig.7A and B). This reduction in responsiveness is not likely to be attributable to a non-specific run-down of the Ca²⁺ response on repeated application of the agonist because, in a separate experimental series, the mean 7,8-DHF-induced Δ[Ca²⁺]_i was unchanged during two successive applications to the same cell (Fig.7D and E; n = 5; P = 0.988). In Nox cells, there was a trend towards a reduction in 7,8-DHF-induced Ca²⁺ responses in the presence of K252a, but this reduction did not reach significance (Fig.7A; n = 6; P = 0.132; Fig.7B). Collectively, these data indicate that the agonist-induced Ca²⁺ response in CHox chromaffin cells was solely due to activation of the full-length TrkB receptor.

Figure 7.

Effects of tyrosine protein kinase inhibition on TrkB receptor-induced intracellular calcium signalling

A and B, specific TrkB agonist (7,8-DHF; 500 nm) induced a significantly greater rise in [Ca²⁺]_i in chronically hypoxic chromaffin cells (2% O₂; n = 14, 156 responsive cells) than normoxic cells (21% O₂; n = 6, 73 responsive cells) (P = 0.0132). The increase in [Ca²⁺]_i in the hypoxic cells was effectively blocked by a tyrosine kinase inhibitor (K252a; 100 nm) (P = 0.031). C, the number of responsive chromaffin cells to the TrkB agonist was significantly greater for hypoxic (n = 14; 374 cells) compared to normoxic (n = 7, 225 cells; P = 0.014) cells. D and E, the Ca²⁺ response was unaltered during two successive applications of 7,8-DHF to the same chronically hypoxic chromaffin cell (n = 5, 81 responsive cells; P = 0.988). Each n represents number of culture dishes sampled. Nox, normoxia; CHox, chronic hypoxia; 7,8-DHF, 7,8-dihydroxyflavone; High K⁺, 30mm KCl. Asterisks denote significant differences between groups (Mann–Whitney analysis and one-way ANOVA). Error bars: ±SEM.

Extracellular calcium influx mediates TrkB receptor signalling in chromaffin cells

TrkB receptor activation can lead to the induction of several downstream signalling cascades, including the phosphorylation of phospholipase Cγ (PLCγ), mitogen-activated protein kinase (MAPK) or phosphoinositide 3-kinase (PI3K). TrkB-mediated IP3 signalling, via activation of PLCγ, leads to a rise in cytosolic Ca²⁺ due to release from intracellular stores such as the endoplasmic reticulum (Poo, 2001). TrkB receptor stimulation has also been associated with the activation of PLC-dependent, non-selective cation channels in the plasma membrane of central neurons, resulting in extracellular Ca²⁺ entry (Li et al. 1999). To determine the source of the Ca²⁺ rise in chromaffin cells following TrkB receptor activation, we first used cyclopiazonic acid (CPA), an inhibitor of the endoplasmic reticulum Ca²⁺ pump (Demaurex et al. 1992), to deplete intracellular Ca²⁺ stores. As illustrated in Fig.8A, pre-treatment with 500 nm CPA failed to inhibit the 7,8-DHF-induced Δ[Ca²⁺]_i in CHox chromaffin cells (n = 6; P = 0.179), suggesting a negligible role for Ca²⁺ from intracellular stores.

Figure 8.

Source of calcium during TrkB receptor signalling in hypoxic chromaffin cells

A, the 7,8-DHF-induced rise in [Ca²⁺]_i was unaffected when cells were pre-treated with CPA (500 nm) an inhibitor of the endoplasmic reticulum Ca²⁺ pump (n = 6, 59 responsive cells; P = 0.179); however, this rise in [Ca²⁺]_i was inhibited by the voltage-gated Ca²⁺ channel blocker, nickel (2 mm; 150 s) (n = 8, 58 responsive cells; P < 0.0001). Each n represents number of culture dishes sampled. 7,8-DHF, 7,8-dihydroxyflavone; K⁺, 30 mm KCl; CPA, cyclopiazonic acid. Asterisks denote significant differences between groups (Mann–Whitney analysis). Error bars: ±SEM.

We next tested whether voltage-gated Ca²⁺ entry contributed to the 7,8-DHF-induced [Ca²⁺]_i response using nickel (2 mm) at a concentration that acts as a general blocker of low and high threshold voltage-gated Ca²⁺ channels (Rich & Rae, 1995), as well as transient receptor potential channels (Ene et al. 2007). In contrast to CPA, 2 mm nickel completely abolished the 7,8-DHF-induced Ca²⁺ responses in CHox chromaffin cells (Fig.8B; n = 8; P<0.0001). As observed in other cell types, there was a rebound in the Ca²⁺ signal soon after the removal of CPA and nickel (Fig.8A and B), which was likely due to non-specific effects of these agents on intracellular Ca²⁺ homeostasis. Taken together, these data suggest that extracellular Ca²⁺ entry, rather than Ca²⁺ release from internal stores, contributes significantly to the elevated [Ca²⁺]_i transients seen during TrkB receptor activation in CHox chromaffin cells.

Major contribution of T-type Ca²⁺ and voltage-gated Na⁺ channels to BDNF-induced Ca²⁺ transients in chronically hypoxic chromaffin cells

TrkB activation of PI3K can also regulate Ca²⁺ entry by increasing the amplitude of Ca²⁺ currents through low threshold Cav3.2 T-type Ca²⁺ channels (Hilaire et al. 2012). Given the established role of Cav3.2 T-type Ca²⁺ calcium channels in fast, low-threshold neurosecretion of catecholamines from chromaffin cells, and their up-regulation during CHox (Carabelli, 2007; Levitsky & Lopez-Barneo, 2009), we investigated a potential link between TrkB signalling and T-type Ca²⁺ channels in CHox chromaffin cells. Interestingly, in the presence of a potent and highly selective T-type Ca²⁺ channel blocker, TTA-P2 (1μm; Choe et al. 2011), the BDNF-induced Ca²⁺ transient was partially blocked by ∼60–80% in some cells (red trace; ∼42% of cells) and completely abolished in others (black trace; ∼58% of cells) (Fig.9A and B; n = 5; P = 0.011). In general, the inhibitory effect of TTA-P2 on the BDNF-induced rise in [Ca²⁺]_i was reversible after wash-out of the drug (Fig.9C and D). When present alone, TTA-P2 had no significant effect on baseline [Ca²⁺]_i levels in CHox chromaffin cells (Fig.9B).

Figure 9.

BDNF-mediated calcium influx is primarily due to T-type calcium channel activation

A and B, BDNF induction of intracellular calcium rise is either partially (∼42% of cells; example red trace) or completely (∼58% of cells; example black trace) inhibited in the presence of T-type channel blocker, TTA-P2 (n = 5, 33 responsive cells; P = 0.011). The average BDNF-induced calcium rise above baseline was significantly greater than in the presence of TTA-P2 (P = 0.019). Calcium levels were not different from baseline in the presence of TTA-P2, and there was no significant difference between [Ca²⁺]_i with or without BDNF in the presence of TTA-P2 (one-way ANOVA). C and D, average [Ca²⁺]_i rises with BDNF application were reduced with blockade of sodium channels with TTX (n = 8; 36 responsive cells; P < 0.0001), and abolished altogether with the subsequent application of TTA-P2 (P < 0.0001; one-way ANOVA). No significant differences in average [Ca²⁺]_i were evident between the first and final BDNF application. High K⁺, 30 mm KCl. BDNF, brain-derived neurotrophic factor; TTA-P2, 3,5-dichloro-N-[1-(2,2-dimethy-tetrahydro-pyran-4-ylmethyl)-4-fluoro-piperidin-4-ylmethyl]-benzamide; TTX, tetrodotoxin. Asterisks denote significant differences between groups or from groups with antagonist application; dagger denotes significant difference from all other groups (one-way ANOVA). Error bars: ±SEM.

In addition to regulating intracellular [Ca²⁺], calcium influx through T-type Ca²⁺ channels can initiate low-threshold spikes, which in turn triggers a burst firing pattern mediated by voltage-gated Na⁺ channels (Perez-Reyes, 2003; Carabelli et al. 2007). To address whether the BDNF-induced Ca²⁺ elevations in CHox chromaffin cells is associated with Na⁺-dependent action potentials we tested the effects of tetrodotoxin (TTX; 1 μm). Blockade of Na⁺ channels with TTX significantly reduced the BDNF-induced Δ[Ca²⁺]_i in CHox cells (Fig.9C and D; n = 8; P < 0.0001). Notably, the smaller but significant Δ[Ca²⁺]_i that persisted in some cases in the presence of TTX was completely abolished by subsequent exposure to TTA-P2 (Fig.9C and D; P < 0.0001). Collectively, these findings demonstrate that the increased intracellular [Ca²⁺] associated with BDNF-TrkB signalling in CHox cells is strongly dependent on Ca²⁺ entry through T-type Ca²⁺ channels. However, TTA-P2 did not completely block the BDNF-induced Δ[Ca²⁺]_i in a subset of CHox cells, suggesting that either the TTA-P2 concentration used did not have a maximal effect on all channels or that additional pathways were involved.

BDNF signalling increased membrane potential and cell excitability of CHox cells

As discussed above, TTX inhibited the BDNF-induced Δ[Ca²⁺]_i in CHox chromaffin cells, suggesting that BDNF increases membrane depolarization and action potential firing. To confirm this directly, we used perforated patch, whole-cell recording to compare the effects of BDNF on the membrane potential in Nox versus CHox cells. Because T-type channel activity is sensitive to the initial membrane potential, only cells with resting potentials more negative than −45 mV were included in the analysis. Application of 100 ng ml⁻¹ BDNF to Nox chromaffin cells from both young and older rats typically produced no detectable response (left panel) or only small membrane depolarizations (<5 mV; right panel; Fig.10A); with action potential firing seen in only a few cases (not shown). By contrast, an enhancement of BDNF responsiveness in CHox cells was observed in both age groups. The effect of BDNF in the CHox cells was dose dependent (n = 4; P < 0.05; Fig.10B) and action potential firing was commonly observed at 50 or 100 ng ml⁻¹ BDNF (Fig. 10B, C and F). The concentration of 100 ng ml⁻¹ BDNF is within the upper physiological range of serum levels seen in humans exposed to hypoxia (15% O₂) for 72 h (Helan et al. 2014). The lowest serum levels (∼20 ng ml⁻¹) observed in hypoxic individuals are equivalent to the average serum levels of BDNF in humans under normoxic conditions. This lower dose was ineffective in CHox cells (Fig.10B and F). As observed for the calcium response, there was a significant increase in the frequency of cells responding to 100 ng ml⁻¹ BDNF from ∼16% in normoxia (4/25 cells; n = 7 dishes) to ∼59% after chronic hypoxia (17/29 cells; n = 7 dishes) (Fig.10D; P = 0.009). In addition, the subset of cells that generated suprathreshold responses during BDNF application had a mean action potential frequency that was significantly higher (∼2×) in CHox (18 cells) compared to Nox cells (4 cells) (Fig.9E; P = 0.029).

Figure 10.

Effects of BDNF on membrane potential of chromaffin cells grown in normoxia versus chronic hypoxia

A, during perforated patch recordings, BDNF (100 ng ml⁻¹) induced either no depolarization (left) or small membrane depolarizations (right) in subpopulations of normoxic (Nox; left) chromaffin cells. B and C, action potential firing was evident in chromaffin cells treated with BDNF following ∼48 h in chronic hypoxia (CHox) and graded increases in BDNF concentrations within the physiological range yielded a dose-dependent increase in spike frequency in chronically hypoxic chromaffin cells (F; n = 6; 20 ng ml⁻¹, 0 Hz; 50 ng ml⁻¹, 2.1 ± 0.6 Hz; 100 ng ml⁻¹, 5.7 ± 0.5 Hz). D, the proportion of BDNF (100 ng ml⁻¹) responsive cells was significantly higher in CHox cells (n = 7, 29 cells) compared to Nox cells (n = 7, 25 cells; P = 0.009). E, in addition, the mean spike frequency in those cells that fired action potentials in response to BDNF was significantly higher in CHox (n = 7, 15 cells) compared to Nox (n = 3; 3 cells) chromaffin cells (P = 0.029). The selective TrkB agonist 7,8-DHF (500 nm) was similarly effective (C). Each n represents number of culture dishes sampled; asterisks denote significant differences between groups (Mann–Whitney analysis). Error bars: ±SEM.

The effects of the selective TrkB receptor agonist (7,8-DHF), on membrane potential appeared to mimic that of BDNF (Fig.10C; right panel). For example, ∼14% of Nox cells (2/14 cells; n = 6 dishes) were responsive to 500 nm 7,8-DHF compared to ∼70% of CHox cells (7/10 cells; n = 3). In addition, mean action potential frequency was ∼2× higher in the CHox cells (1.21 ± 0.29 Hz) compared to Nox cells (0.68 ± 0.19 Hz) in response to the specific agonist.

Discussion

In this study, we provide evidence that chronic hypoxia upregulates the expression of the neurotrophin receptor TrkB in adrenomedullary chromaffin cells (AMCs). Using amperometry, ratiometric intracellular Ca²⁺ imaging and electrophysiological recording, we further demonstrate that BDNF-TrkB signalling in chronically hypoxic AMCs leads to enhanced catecholamine secretion via an increase in membrane excitability and Ca²⁺ influx via voltage-gated T-type calcium channels. Similar responses were observed in AMCs isolated from both young (10-day-old) and mature (3- to 4-week-old) rats, suggesting that neural innervation of chromaffin cells (which develops between first and second week postnatal) was not a confounding factor. The enhancement of catecholamine secretion was mainly attributable to an increase in frequency of quantal events, with no significant change in mean quantal charge. This pattern was consistent with previous reports of K⁺-induced mild depolarizations of chronically hypoxic adult chromaffin cells, which also produced increases in quantal secretion frequency but not quantal content (Carabelli, 2007). In addition, we found the specific TrkB agonist 7,8-DHF (Jang et al. 2010) stimulated a rise in intracellular Ca²⁺ and catecholamine secretion, signifying that neither the low affinity p75^NTR receptors nor other Trk receptors were involved. Moreover, inhibition of the 7,8-DHF-induced rise in Ca²⁺ by the tyrosine kinase inhibitor (K252a) indicated the response was mediated by the full-length TrkB rather than by the short, tyrosine kinase-deficient TrkBT1 isoform. Collectively, the present study highlights a novel paracrine mechanism that contributes to the remodelling of chromaffin cell stimulus–secretion coupling during chronic hypoxia.

The present study also implicated a role for hypoxia inducible factor 2α (HIF-2α) in BDNF-TrkB signalling in chromaffin cells during chronic hypoxia. HIFs, including HIF-1α, -2α and -3α, are heterodimers that consist of constitutively expressed α and β subunits. The HIF α subunit is rapidly degraded in normoxic conditions; therefore, it is unable to bind to the β subunit so as to initiate transcription of specific gene targets (Wang et al. 1995; Huang et al. 1998). Although expression of HIF-1α is widespread, HIF-2α is expressed in a more restricted pattern and known to play a predominant regulatory role in cells of the sympathoadrenal lineage (Bishop et al. 2008; Semenza, 2012; Richter et al. 2013). In fact, unlike the transient expression of HIF-1α, HIF-2α expression is upregulated within 2 h and remains elevated for at least 24 h in an immortalized adrenal chromaffin cell line, i.e. MAH cells (Brown & Nurse, 2008; Brown et al. 2009). Using MAH cells we show that, while upregulation of TrkB mRNA occurs in scrambled control MAH cells, it is completely prevented in HIF-2α-deficient (>90% knockdown; Brown & Nurse, 2008) MAH cells. The HIF transcription factors induce transcriptional regulation by interacting with hypoxia response elements (HREs) within the promoter region of the target genes (Semenza, 2012). Notably, in neuroblastoma cells many HREs have been identified in regulatory active domains of the promoter sequence of the TrkB gene (NTRK2); another member of the HIF family, HIF-1α, binds directly to the TrkB promoter in these cells (Martens et al. 2007). The presence of HREs in the NTRK2 promoter sequence and the prevention of TrkB upregulation in HIF-2α-deficient MAH cells suggest that TrkB is a novel target of HIF-2α in chromaffin cells during chronic hypoxia.

TrkB-induced modulation of chromaffin cell function in chronic hypoxia

In general, an increase in efficiency of stimulus–secretion coupling by chromaffin cells is a hallmark of the physiological response of the mammalian adrenal medulla to chronic stress. Multiple intracellular and intercellular pathways are potential targets for chromaffin cell remodelling during adaptation to stress, including modifications in membrane excitability and Ca²⁺ signalling (Guérineau et al. 2012). For example, an upregulation of low-threshold α_1H T-type Ca²⁺ channels has been reported in adult rat chromaffin cells exposed to chronic hypoxia in vitro (3% O₂; 12–18 h) (Carabelli et al. 2007), conditions similar to the present study. Since a significant number of these T-type channels are open at negative membrane potentials near rest, they contribute to low-threshold catecholamine (CAT) release in response to mild depolarizations (Giancippoli et al. 2006; Carabelli, 2007; Carabelli et al. 2007). In our study, acute BDNF application to CHox chromaffin cells led to CAT secretion, attributable largely to activation of T-type Ca²⁺ channels, action potential firing, and entry of extracellular Ca²⁺. In particular, only a fraction of the BDNF-evoked Δ[Ca²⁺]_i was reduced in the presence of TTX, but the remainder was eliminated by the specific T-type Ca²⁺ channel blocker, TTA-P2, in most cases. These data suggest that while TTX-sensitive voltage-gated sodium channels contribute to the Δ[Ca²⁺]_i signal, likely via activation of high threshold L-type channels, it appears that T-type Ca²⁺ channels played a major role in initiating the response to BDNF near the resting potential. In pyramidal neurons, TrkB-mediated depolarization has been linked to the activation and opening of both sodium and calcium channels on postsynaptic pyramidal neurons (Blum et al. 2002; Lang et al. 2007; Amaral & Pozzo-Miller, 2012). Inward current via the voltage-gated TTX-insensitive sodium channels, Na_v1.9, is evoked within milliseconds of BDNF interaction with TrkB in pyramidal neurons (Blum et al. 2002); however, Nav1.9 appears to be absent from rodent chromaffin cells (Vandael et al. 2015). Previous studies on neonatal rat pontine neurons demonstrated that PLC-mediated cation influx via TRPC3 channels might also occur following TrkB activation (Li et al. 1999). However, further studies are required to determine whether TRPC3 channels in CHox chromaffin cells contribute to the residual BDNF-evoked Δ[Ca²⁺]_i in the few cases where it persisted in the presence of TTA-P2.

Remodelling associated with CHox-induced T-type Ca²⁺ channel expression and cellular distribution in rat chromaffin cells has been shown to be linearly related to changes in the rate of vesicular release, rather than quantal size (Carbone et al. 2006). This parallels the changes in vesicular release elicited by BDNF in the present study. In normoxic conditions, the expression of T-type calcium channels is highly restricted in adult rat chromaffin cells (Hollins & Ikeda, 1996; Carabelli, 2007), but is upregulated via HIF-2α during hypoxia (Del Toro et al. 2003). Similarly, we showed that during hypoxia TrkB is also upregulated in chromaffin cells via HIF-2α. Interestingly, stabilization and nuclear translocation of HIF proteins via activation of PLC, a downstream target of TrkB, is required for the recruitment of the T-channel subtype Cav3.2 during hypoxia in a PC12 cell line (Yuan et al. 2008). Another possible link between Cav3.2 and TrkB was found in D-hair sensory neurons, which express high levels of both Cav3.2 and TrkB. Deprivation of NT-4 and the subsequent inhibition of TrkB led to significant reductions in T-current amplitude, an effect rescued by restoring NT-4 levels (Hilaire et al. 2012). While in these neurons, TrkB activation had no transcriptional effect on Cav3.2, inhibition of PI3K activation via TrkB caused significant reductions to calcium current through the low threshold T-type calcium channels. These results suggest that downstream targets of TrkB are responsible for post-translational modification of the calcium channels, and in this way, the T-type channels undergo an increase in recruitment, activity, or both. The systematic inhibition of selective downstream targets of TrkB signalling, and the subsequent effect on T-type calcium channel activity, will help elucidate the molecular mechanisms underlying BDNF-induced cell excitability and Ca²⁺ influx in chromaffin cells during hypoxic stress.

In the central nervous system (CNS), there is precedence for a role of BDNF-TrkB signalling in modulating synaptic plasticity. For example, BDNF-induced vesicular release of glutamate has been demonstrated in cultured post-natal hippocampal neurons (Lessmann et al. 1994), embryonic hippocampal cultures (Levine et al. 1995), and adult hippocampal slices (Scharfman, 1997). In other CNS tissues, BDNF also potentiates the synaptic release of several other neurotransmitters, such as monoamines, from neurons in the hypothalamus (Rodriguez Fermepin et al. 2009), the corpus striatum (Goggi et al. 2002) and visual cortex (Akaneya et al. 1997; Carmignoto et al. 1997). The mechanism by which TrkB activation regulates synaptic output in these neurons is not fully understood but several multifaceted roles have been proposed. TrkB-mediated effects in CNS neurons have included the upregulation of synaptic docking machinery (Pozzo-Miller et al. 1999; Jovanovic et al. 2000), potentiation of intracellular Ca²⁺ influx (Kang & Schuman, 1995; Amaral & Pozzo-Miller, 2012), and modulation of cation channel expression and/or activity (Hilborn et al. 1998; Urbano & Buno, 2000). The present study shows that TrkB activation also induces intracellular Ca²⁺ influx in AMCs, especially after chronic hypoxia, and that the pathway involves enhanced activation of T-type calcium channels.

Paracrine stimulation of chromaffin cells via BDNF-TrkB signalling?

In this study, we examined not only the expression of TrkB within the adrenal gland but also that of the TrkB ligand, BDNF. In contrast to the pattern of TrkB expression, we found BDNF was highly expressed within the rat adrenal cortex, consistent with previous studies (Suter-Crazzolara et al. 1996; Schober et al. 1999; Szekeres et al. 2010). Compared to the cortex, expression of BDNF was significantly lower in the medullary region, where its location was predominantly associated with microvascular cells as previously reported (Wang et al. 2006). Taken together, BDNF and TrkB expression within the adrenal gland is ideally situated for both local (via adrenal cortex) and systemic (via vascular) paracrine signalling.

Given that chronic hypoxia results in the upregulation of TrkB in chromaffin cells, it is likely this paracrine mechanism contributes to the enhanced adrenal catecholamine secretion seen during whole animal exposures to chronic hypoxia in vivo (Cannon & Hoskins, 1911; Johnson et al. 1983). It is noteworthy that in previous reports, immobilization stress (60 min) caused an increase in TrkB mRNA expression in the adrenal medulla of adult rats (Kondo et al. 2010), as well as an increase in both plasma BDNF concentration and BDNF-induced adrenal catecholamine secretion (Kondo et al. 2012). Similar changes to BDNF levels have been noted during hypoxia. In adult humans, a 30 min or 72 h exposure to hypoxia caused an increase in circulating plasma BDNF levels (Hubold et al. 2009; Helan et al. 2014), and a similar increase occurred in infants during hypoxic episodes associated with birth (Nikolaou et al. 2006). Importantly, the upregulation of TrkB in chromaffin cells, combined with the increase in circulating BDNF, during hypoxic stress provide a non-neurogenic pathway for enhancing catecholamine secretion.

In conclusion, the findings presented in this study clearly demonstrate the important role of BDNF via TrkB in the modulation of catecholamine release by chromaffin cells following hypoxia exposure, and revealed the integrative cellular mechanisms underlying this HIF-dependent effect. In doing so, this work has highlighted a novel regulatory signalling pathway that may be involved in activating and potentiating critical endocrine responses within the adrenal medulla during hypoxic stress. Further elucidation of BDNF-TrkB signalling using in vivo models will help to unravel the significance of the TrkB-mediated catecholecholamine release to the sympathoadrenal system during adaptation to environmental stress.

Glossary

AMC

adrenomedullary chromaffin cell

BDNF

brain-derived neurotrophic factor

CAT

catecholamines

CHox

chronic hypoxia

DHF

dihydroxyflavone

HIF

hypoxia-inducible factor

MAH

immortalized v-myc, adrenal-derived

HNK1⁺

Nox, normoxia

TrkB

tropomyosin-related kinase B

TTA-P2

3,5-dichloro-N-[1-(2,2-dimethy-tetrahydro-pyran-4-ylmethyl)-4-fluoro-piperidin-4-ylmethyl]-benzamide; TTX tetrodotoxin

Additional information

Competing interests

The authors declare no competing financial interests.

Author contributions

A.L.S. prepared all cultures, collected data and analysed all experiments pertaining to the immunohistochemistry, RT-qPCR, Ca²⁺ imaging and amperometry data. M.Z. was responsible for collection and analysis of the electrophysiological data. A.L.S. and C.A.N. planned and designed all the experiments. A.L.S. prepared the figures and wrote the manuscript, C.A.N. edited the manuscript, and all authors approved the final version of the manuscript.

Funding

This work was funded by a Natural Sciences and Engineering Research Council (NSERC) Discovery Grant to C.A.N. A.L.S. is a NSERC Post-Doctoral Fellow.