Stimulation of bursting in pre-Bötzinger neurons by Epac through calcium release and modulation of TRPM4 and K-ATP channels


  • Sergej L. Mironov,

    1. DFG-Center of Molecular Physiology of the Brain, Department of Neuro- and Sensory Physiology, Georg-August-University, Göttingen, Germany
    Search for more papers by this author
  • Ekaterina Y. Skorova

    1. DFG-Center of Molecular Physiology of the Brain, Department of Neuro- and Sensory Physiology, Georg-August-University, Göttingen, Germany
    Search for more papers by this author

Address correspondence and reprint requests to S. L. Mironov, DFG-Center of Molecular Physiology of the Brain, Department of Neuro- and Sensory Physiology, Humboldtallee 23, 37073 Göttingen, Germany. E-mail:


J. Neurochem. (2011) 117, 295–308.


The exchange factor directly activated by cAMP (Epac) can couple cAMP production to the activation of particular membrane and cytoplasmic targets. Using patch-clamp recordings and calcium imaging in organotypic brainstem slices, we examined the role of Epac in pre-Bötzinger complex, an essential part of the respiratory network. The selective agonist 8-(4-chlorophenylthio)-2′-O-methyl-cAMP (8-pCPT) sensitized calcium mobilisation from inositol-1,4,5-trisphosphate-sensitive internal stores that stimulated TRPM4 (transient receptor potential cation channel, subfamily M, Melastatin) channels and potentiated the bursts of action potentials. 8-pCPT actions were abolished after inhibition of phospholipase C with U73122 and depletion of calcium stores with thapsigargin. Caffeine-sensitive release channels were not modulated by 8-pCPT. Epac inhibited ATP-sensitive K+ channels that also led to the enhancement of bursting by 8-pCPT. Bursting activity, spontaneous calcium transients and activity of TRPM4 and ATP-sensitive K+ channels were potentiated after brief exposures to bradykinin and incubation with wortmannin produced opposite effects that can be explained by changes in phosphatidylinositol 4,5-bisphosphate levels. 8-pCPT stimulated the respiratory motor output in functionally intact preparations and the effects of bradykinin and wortmannin were identical to those observed in organotypic slices. The data thus indicate a novel pathway of controlling bursting activity in pre-Bötzinger complex neurons through Epac that can involved in reinforcement of the respiratory activity by cAMP.

Abbreviations used

8-(4-chlorophenylthio)-2′-O-methyladenosine-3′,5′-cyclic monophosphate


artificial CSF




action potentials




endoplasmic reticulum


fluorescence resonance energy transfer




ATP-sensitive K+-channels


postnatal day


phosphatidylinositol 4,5-bisphosphate


protein kinase A


phospholipase C


pre-Bötzinger complex


transient receptor potential cation channel, subfamily M (Melastatin)

Epac is an acronym for the exchange proteins activated directly by cAMP, a family of cAMP-regulated guanine nucleotide exchange factors. By serving as a cAMP-binding protein with guanine nucleotide exchange factor activity, Epac couples cAMP production to the activation of small G-proteins, microtubule-associated and secretory proteins as well as ion channels and transporters. Thus, the actions of cAMP, which at one time were thought to be mediated exclusively by protein kinase A (PKA), must now to be re-evaluated in the light of recent findings that indicate an important role for Epac in cell physiology (Bos 2003; Seino and Shibasaki 2005; Holz et al. 2006). Identification of previously unrecognized physiological processes regulated by Epac became possible by the development of Epac-selective cAMP analogues such as 8-(4-chlorophenylthio)-adenosine-3′,5′-cyclic monophosphate (8-pCPT) which activates Epac proteins but fails to activate PKA. This agonist has been widely used in recent years to reveal the role of Epac in the cAMP-dependent regulation of ion channel function, intracellular calcium signalling, ion transporter activity and exocytosis. The characterization of a novel cAMP signal transduction mechanism reinvigorated interest in cyclic nucleotide research. Most of the new data has been obtained in excitable heart cells, pancreatic β-cells and non-excitable cell lines. Recent studies (Novara et al. 2004; Zhong and Zucker 2005; Ma et al. 2009; Ster et al. 2009) indicate a potential role of Epac in the nervous system, especially in synaptic transmission and plasticity.

We examined Epac-mediated regulation of TRPM4 (transient receptor potential cation channel, subfamily M, Melastatin) and K-ATP (ATP-sensitive K+-channels) ion channels in pre-Bötzinger complex (preBötC), a brainstem region that is thought to contain a kernel structure of the respiratory network generating inspiratory rhythmic activity (reviewed in Feldman and Del Negro 2006). Smith et al. (1991) were the first to show that the neuronal network in preBötC produces an autonomous respiratory-like output that is sent to hypoglossal (XII) nucleus and can be recorded from its rootlets. Subsequent studies revealed a plethora of possibilities to modulate the output of preBötC network through modifications in the activity of single neurons and synaptic interactions. Particularly interesting is cAMP which elevation reinforces the rhythm (Lalley et al. 1997; Mironov et al. 1999; Ruangkittisakul et al. 2006) that can be beneficial in treatment of various respiratory disorders (Richter et al. 2000). A recent discovery of new cAMP effector raises a question which role Epac plays in cAMP-dependent mechanisms of modulation of network activities and the types of ion channels involved.

Using calcium imaging and single-channel patch-clamp recordings we examined the effects of Epac on neuronal activity of preBötC neurons in organotypic brainstem slices (Hartelt et al. 2008). Electrophysiology was combined with functional calcium imaging in neurons expressing fluorescence-resonance-energy-transfer (FRET)-based calcium sensor D3cpv (Mironov et al. 2009a). Many preBötC neurons showed intrinsic bursting activity and repetitive calcium transients which both were stimulated by Epac agonist 8-pCPT. It sensitized calcium release from inositol-1,4,5-trisphosphate (IP3)-sensitive internal stores that potentiated the activity of TRPM4 channels. 8-pCPT suppressed K-ATP channels that additionally enhanced bursting. Activities of TRPM4 and K-ATP increased after brief exposure to bradykinin and decreased after incubation with wortmannin, the treatments often used to manipulate phosphatidylinositol 4,5-bisphosphate (PIP2) levels (Loew 2007; Pian et al. 2007). All effects of Epac stimulation were reproduced in acutely isolated functionally intact preparation. Our data thus indicate a novel pathway of cAMP signalling in preBötC neurons that is independent of protein kinase A and can be involved in beneficial effects of cAMP in the respiratory network.



All animals were housed, cared for and killed in accordance with the recommendations of European Commission (No. L358, ISSN 0378-6978), and protocols were approved by the Committee for Animal Research, Göttingen University. Organotypic brainstem slices containing preBötC were obtained from mice (Naval Medical Research Institute, postnatal day 3–5) as described previously (Hartelt et al. 2008; Mironov et al. 2009a,b). Animals were prepared at postnatal day P3 and slices were placed on support membranes (Millicell-CM Inserts, PICMORG50; Millipore Corporation, Bedford, MA, USA). Preparations were examined 1–3 weeks later and each test was repeated with at least four different preparations. Statistical analyses were performed using Statview program (version 5.0.1, SAS Institute Inc., NC, USA). Mean values ± SEM are reported for all essential experiments which number and statistical significance (p) are given in brackets.

Slices in the experiments were fixed on a coverslip in the recording chamber and continuously superfused at 34°C with artificial CSF (ACSF) containing (in mM) 136 NaCl, 5 KCl, 1.25 CaCl2, 0.8 MgSO4, 0.4 NaH2PO4, 0.3 K2HPO4, 3.3 NaHCO3, 6 glucose, pH 7.4 and saturated with 95% O2–5% CO2. All chemicals were from Sigma (Deisenhofen, Germany) and all cAMP-related drugs were from Biolog (Hamburg, Germany). cAMP changes were elicited by applying 45 mM K+, 0.1 mM ATP, 1 μM forskolin to stimulate adenylate cyclase and 1 μM rolipram to inhibit phosphodiesterase PDE4. High-K+ solution was prepared by exchanging Na+ in ACSF. Other testing solutions were produced by adding aliquots of corresponding stock solutions directly to ACSF.

‘Rhythmic’ slices (thickness, 600 μm) containing a functional respiratory network (Smith et al. 1991) were prepared from brainstem of P5–P7 mice (Naval Medical Research Institute) as described previously (Mironov et al. 1998). After preparation, the slice was placed in the recording chamber on a nylon mesh, overlaid with a threaded home-made platinum-iridium ‘horse-shoe’ for mechanical stability and the neurons were visualized with infrared differential interference contrast illumination. Thirty minutes after the slice was positioned in the chamber, one of the hypoglossal (XII) rootlets was sucked into a blunt electrode for extracellular recordings of the respiratory motor output. A robust repetitive activity was established after elevation of extracellular K+ to 8 mM and remained stable for 6–12 h.

Calcium imaging

The measurements were made as described previously (Mironov et al. 2009a). Briefly, we expressed FRET-based calcium sensor protein D3cpv in slices using recombinant adeno-associated virus vector containing synapsin 1 gene promoter (Hartelt et al. 2008). The slices in experiments were placed in a chamber mounted on an upright microscope (Axioscope 2, Carl Zeiss AG, Göttingen, Germany) and visualized with a 40× objective. The excitation light (425 nm) from CoolLED (BFI Optilas, Puchheim, Germany) was attenuated to 10–30%. Changes in fluorescence emission intensities at 535 nm (FRET acceptor) and 470 nm (enhanced cyan fluorescent protein, ECFP; FRET donor) were measured after separation of signals with Optosplit (BFI Optilas) through dichroic mirror (495 nm) and additional filtering at 470 ± 12 and 535 ± 15 nm. Images were captured by cooled CCD camera (BFI Optilas) using ANDOR software (500 × 500 pixels at 12 bit resolution). Fluorescence signals were analyzed offline with MetaMorph software (Universal Imaging Corp., Downington, PA, USA) and custom-made programs. Calcium levels were obtained from the soma from ratios of FRET and CFP signals as described previously (Mironov et al. 2009a).

Figure 1(a) presents the image of the transduced brainstem slice at low magnification and Fig. 1(b) shows individual neurons in preBötC. Large-scale calcium imaging revealed regular activity in preBötC area (Fig. 1c) which disappeared after blockade of excitatory transmission with α-amino-3-hydroxy-5-methylisoxazole-4-propionate (AMPA) antagonist 6-cyano-7-nitroquinoxaline-2,3-dione (CNQX) (Fig. 1d). The same effect was observed in functionally intact acute slices (Fig. 1e).

Figure 1.

 preBötC network and its activity. (a) Organotypic brainstem slice transduced with FRET-based neuron-targeted calcium sensor D3cpv. The image was constructed from individual overlapping frames collected in the same focal plane with ×10 objective. Position of pre-Bötzinger complex is indicated in the right half of the slice. (b) preBötC neurons at higher magnification (×40 objective). (c) Spontaneous activity measured in preBötC area (160 × 160 μm). Each trace shows calcium changes averaged in quadrants (40 × 40 μm) which contained from four to six neurons. (d, e) Suppression of rhythmic activity by AMPA antagonist CNQX (10 μM) in organotypic and acute slices. The data present calcium changes averaged over preBötC area in organotypic slices (d) and integrated motor output in the functional slice recorded from hypoglossal nerve (∫XII, e).

Electrophysiology and single channel analysis

The activity of preBötC neurons was recorded in cell-attached mode. The patch electrodes (resistance 3 MOhm) were made from borosilicate glass (Clark Instruments, Pangbourne, UK). The membrane currents were measured using the amplifier EPC-7 (HEKA Elektronik, Lambrecht/Pfalz, Germany) as described previously (Mironov et al. 1998). They were filtered at 3 kHz (−3 dB), digitized at 10 kHz, and stored for offline analysis. The pipette solution in all recordings contained (in mM): 125 K+-gluconate, 10 NaCl, 2 MgCl2, 10 HEPES, 0.5 Na2ATP, pH adjusted to 7.4 with KOH. Assuming that intracellular K+ concentration is 145 mM, the equilibrium potential for K+ across the patch should be −4 mV, close to the reversal potential (0 mV) of non-selective TRPM4 channels which are equally permeable for Na+ and K+. Using this estimate we assessed the resting potential of neurons from the reversal potential of single TRPM4 channel currents in cell-attached recordings. The mean value was −55 ± 5 mV (= 12), close to that observed in preBötC neurons in functionally intact brainstem slices (Mironov et al. 1999).

To patch neurons, we positioned the pipette above the slice directly over the center of fluorescent soma of particular neuron and then entered the slice strictly vertical until a contact with a target was established as indicated by slight changes in neuron appearance in both fluorescence and infrared differential interference contrast images. The gigaseal was formed and, in the case of bursting activity, a temporal correlation of spikes and calcium transients additionally verified that calcium and patch-clamp measurements were made from the same cell. In recordings of initially quiescent cells, a similar correlation between bursting and calcium activities was observed after 8-pCPT application (see below).

TRPM4 or K-ATP channels were identified by their conductances (25 pS for TRPM4 channels, Fig. 2a; and 75 pS for K-ATP channels, Fig. 7a). K-ATP channels in preBötC neurons are composed of Kir.6. 1 and SUR2 subunits (Haller et al. 2001). The molecular identity of TRPM4 channels has to be established, but their characteristics in the acute (Mironov 2008) and organotypic slices presume the same biophysical and pharmacological properties as their counterparts in non-excitable cells (Nilius et al. 2007). The conductances were measured from the slopes of single-channel currents during voltage ramps (100 mV for 1 s). During the first 5 min after obtaining gigaseal the protocol was repetitively applied to verify the presence of only one channel type in the patch. When two different conductance levels (25 and 75 pS) were noticed, the patch was discarded. This test was also applied intermittently during the experiment and at the end of it. Identification of TRPM4 channels was confirmed by their blockade by flufenamate and Gd3+ as exemplified in Fig. 2(b). K-ATP channels were blocked by glibenclamide (100 μM in cell-attached and 30 μM in inside-out recordings) and potentiated by diazoxide (100 μM in cell-attached and 30 μM in inside-out recordings) as exemplified in Fig. 7(b). The effects of the drugs (> 4 for each agent tested) were identical to those described previously in preBötC inspiratory neurons (Mironov et al. 1998; Mironov 2008).

Figure 2.

 TRPM4-channels in preBötC and the effects of Epac agonist. (a) Original patch-clamp recording (upper trace) shows action potential currents (upwardly directed spikes) and openings of single TRPM4 channels (downwardly directed deflections, the closed and two open levels are indicated by dashed lines). The lower trace was obtained after subtraction of action potential currents as described previously (Mironov 2008). Voltage-dependence of single TRPM4 channels is shown on the right. The points give the values of mean channel current ± SEM at different patch potentials and insets show the episodes of single channel activity. Closed and two open states are indicated by arrowheads. A straight line (slope 25 ± 2 pS, = 12) approximates single channel conductance. (b) Pharmacology of TRPM4 channels. Cell-attached recordings were made in two different neurons in a control (upper traces) and 10 min after addition of 300 μM flufenamate to the bath (left) or in the presence of 30 μM Gd3+ in the pipette (right) (lower traces). Note disappearance of bursting and channel activities in the presence of flufenamate and bursting activity recorded with Gd3+-containing pipette without TRPM4 channel activity in the patch. (c, d) Potentiation of intracellular calcium transients and TRPM4 channel activity by the Epac agonist. Application of 0.3 μM 8-pCPT induced bursting in a quiescent neuron (c) and potentiated it in rhythmically active cell (d). The four traces present original recordings of membrane current (Im); single channel activities after subtraction of APs, changes in open state probability (popen), and intracellular calcium fluctuations in the soma, respectively. The graphs on the right show mean changes in interburst intervals during application of 0.3 μM 8-CPT.

Figure 7.

 Epac inhibition of K-ATP channels. (a) I–V relationship. The points indicate mean single channel currents (± SEM) and insets show typical activity patterns (the open and closed states are indicated by ‘o’ and ‘c’ arrowheads, respectively). (b) Pharmacology of K-ATP channels. Presented are cell-attached recordings in a control and 4 min after addition of blocker glibenclamide and activator diazoxide to the bath. (c) Inhibition of K-ATP channels and enhancement of bursting activity by 8-pCPT. (d) Suppression of K-ATP channel openings by forskolin and their further inhibition by 8-pCPT.

Open state probabilities (popen) of TRPM4 and K-ATP channels showed periodic fluctuations in relation to bursting activity. For TRPM4 channels popen increased shortly before and during the burst of action potentials (APs) whereas the activity of K-ATP channels increased during and shortly after the burst. Previously we showed that activation of TRPM4 channels is responsible for depolarisation drive that triggers the burst of APs (Mironov 2008) and transient increases in the activity of K-ATP channels are caused by depletion of ATP levels during the burst (Haller et al. 2001). For the analysis of channel activity APs were subtracted from the original traces (Fig. 2a) as described previously (Mironov 2008). The time-dependent changes in open state probability were calculated as a ratio of the mean current in the sliding window (50 ms duration) and the unitary current. Periodic changes in the activity of the channels were characterized by their maximal and minimal values which were obtained by averaging popen values for 10 peaks and troughs in a control and after the effects of drugs reached a steady state.


Modulation of bursting and TRPM4 channel activities by 8-pCPT

8-(4-Chlorophenylthio)-2′-O-methyladenosine-3′,5′-cyclic monophosphate has high specificity to Epac and corresponding EC50 = 40 nM (Dao et al. 2006) is much lower than EC50 = 2.8 μM for PKA activation (Dao et al. 2006). We applied the agonist at 0.3 μM that was in the range of concentrations used in other studies (Kang et al. 2003, 2008; Ster et al. 2009; Ma et al. 2009). At this concentration 8-pCPT should activate Epac to 90% and PKA to only 9%.

Pre-Bötzinger complex neurons were either quiescent (55 of 128 neurons recorded or 43%) or showed repetitive bursting activity (73/128 = 57%, the data collected in 64 cultured slices). The activities were thus distributed in similar proportion as in the functionally intact brainstem slices (Mironov 2008). 8-pCPT induced rhythmic activity in quiescent cells (Fig. 2c) and enhanced intrinsic bursting in the ‘pacemakers’ (Fig. 2d). Three minutes after application of 0.3 μM 8-pCPT the mean number of APs in the burst was 9.6 ± 2.1 vs. 4.2 ± 1.1 in a control (< 0.01) and the interburst interval was 6.2 ± 0.3 s vs. 10.2 ± 0.3 s, respectively (= 12, < 0.01). The changes in bursting activity correlated well with the increases in the activity of calcium-dependent TRPM4 channels. popen values in bursting neurons fluctuated between 0.05 ± 0.01 and 0.25 ± 0.02 in a control and after 3 min with 8-pCPT the minimal and maximal values increased to 0.20 ± 0.03 and 0.51 ± 0.04, respectively (= 12, < 0.01).

As a negative control, we used membrane-permeant analogue 8-(4-chlorophenylthio)-2′-O-methylxanthosine-3′,5′-cyclic monophosphate which does not activate Epac. Its application at 1 μM for 10 min did not change ion channel and bursting activities (Fig. 3a, six neurons examined in three different slices). We also used Brefeldin A which is often applied to inhibit Epac signalling (Rocher et al. 2009; Ster et al. 2009). After incubation of slices with 25 μM Brefeldin A for 30 min the calcium transients, bursting and TRPM4 channel activities were depressed (Fig. 3b). The observed decrease in the frequency of bursts from 6.4 ± 0.5 to 4.1 ± 0.3 per minute (= 4, < 0.05) can indicate a tonic activation of Epac at rest. Subsequent application of 8-pCPT was without effect and the bursting frequency was 4.4 ± 0.3 per minute (> 0.05).

Figure 3.

 Specificity of 8-pCPT actions in preBötC neurons. (a) Absence of the effects of non-active 8-(4-chlorophenylthio)-2′-O-methylxanthosine-3′,5′-cyclic monophosphate (8-pCPT-2′-O-Me-cXMP) used as a negative control of Epac actions. (b) Brefeldin A (100 μM for 20 min) inhibited bursting and calcium activities and occluded the effects of 8-pCPT. (c) Blockade of bursting activity after bath application of TRPM4 blocker Gd3+ (30 μM). Because spontaneous calcium transients are based on calcium release from internal stores, they were not affected by Gd3+ in the bath and fluctuations of TRPM4 channels within the patch were maintained. Subsequent application of 0.3 μM 8-pCPT potentiated calcium transients, stimulated TRPM4 channels in the patch but did not induce bursting activity. (d) Abolishment of bursting after blockade of excitatory glutamatergic transmission with AMPA antagonist CNQX (10 μM). TRPM4 channel activity was decreased and then potentiated by 8-pCPT without appearance of spontaneous calcium transients and bursting activity.

Bath application of 30 μM Gd3+, a blocker of TRPM4 channels (Mironov 2008), abolished bursting (Fig. 3c, n = 6). Repetitive calcium transients and periodic activation of TRPM4 channels in cell-attached patches were yet present, because the patch was spatially isolated from Gd3+-containing solution in the bath. 0.3 μM 8-pCPT applied in the presence of Gd3+ increased the amplitude of calcium transients and open state probability of TRPM4 channels to the levels observed without Gd3+ in the bath. Bursting activity was not restored, because it requires activation of TRPM4 channels in the membrane of the whole cell that was blocked by Gd3+.

The bursting was abolished by AMPA antagonist CNQX (10 μM, Fig. 3d). Calcium transients disappeared, the activity of TRPM4 channels diminished and showed no fluctuations. This indicates an importance of excitatory glutamatergic transmission in generation of rhythmic bursting activity in organotypic slices, similar to that established in the functional slice preparation (Koshiya and Smith 1999). In the presence of CNQX, 0.3 μM 8-pCPT increased popen from 0.04 ± 0.01 to 0.25 ± 0.02 (n = 4, p < 0.01) but spontaneous activity did not appear.

The activator of adenylate cyclase forskolin (1 μM) potentiated TRPM4 channels that reinforced bursting. The effect stabilized after 2 min and subsequent application of 8-pCPT enhanced the activities further (Fig. 4a). The rightmost graph in Fig. 4(a) shows mean increases in bursting frequency and number of APs in the bursts in six experiments. The maximal open state probability of TRPM4 channels during the bursts was 0.22 ± 0.04 in a control, increased 3 min after beginning forskolin application to 0.32 ± 0.04 (< 0.05) and further to 0.48 ± 0.06 (< 0.05) 3 min after subsequent addition of 0.3 μM 8-pCPT. Similar effects were obtained with 1 μM rolipram, a phosphodiesterase (PDE4) inhibitor (= 6, data not shown). When 8-pCPT was applied first, a stimulatory action of forskolin was not observed (Fig. 4b). After treatment of slices with PKA inhibitor (N-[2-(p-bromocinnamylamino)ethyl]-5-isoquinoline sulfonamide hydrochloride, protein kinase A inhibitor (H-89; 10 μM, 10 min), the effects of forskolin and 8-pCPT were not modified (Fig. 4c). Direct activation of PKA with phosphodiesterase-resistant and membrane-permeable agonist N6-benzoyladenosine-3′,5′-cyclic monophosphorothioate (10 μM for 10 min) did not modify bursting (Fig. 4d). The mean frequency and the number of APs in the burst were 4.2 ± 0.4 vs. 4.4 ± 0.5 per minute and 3.1 ± 0.3 vs. 3.6 ± 0.4, respectively (> 0.05 for both sets of the means). Application of 0.3 μM 8-pCPT (Fig. 4d) increased these values to 8.2 ± 0.6 per minute and 7.1 ± 0.5 (= 5, < 0.01).

Figure 4.

 cAMP, Epac and PKA effects on bursting and TRPM4 channel activities. (a) Potentiation of bursting with forskolin and its further enhancement by 8-pCPT. Original recordings of membrane current, activity of TRPM4 channels and calcium fluctuations are presented on the left and the graph on the right shows changes in interburst interval and number of spikes in bursts. Forskolin potentiated calcium transients, TRPM channels and bursting that was further enhanced by 8-pCPT. (b) Occlusion of forskolin effects by 8-pCPT. (c) PKA inhibitor H-89 (10 μM for 10 min) did not prevent the effects of forskolin and 8-pCPT. (d) Direct activation of PKA with Sp-cAMP did not affect the rhythmic activity and the effects of 8-pCPT. The traces shown in (b–d) were recorded in a control and 3 min after beginning the applications.

8-pCPT effects on calcium handling

Bursting activity of inspiratory neurons in acute brainstem slices is interrupted by brief membrane depolarisations (Mironov 2008). Figure 5(a) shows that high-K+ (45 mM for 10 s) induced sustained firing that outlasted the stimulus and correlated with transient calcium increases and potentiation of TRPM4 channels. After calcium recovery to resting levels the neuronal activity was yet depressed and then the bursting was slowly restored after appearance of fluctuations in the activity of TRPM4 channels. High-K+ increased calcium from 0.06 ± 0.02 μM to 0.25 ± 0.03 μM and it then decayed with a mean time-constant 12.4 ± 0.3 per second (= 6). In the presence of 8-pCPT, the depolarisation-induced calcium increases and decay time-constants were bigger (0.32 ± 0.03 μM and 15.6 ± 0.3 per second, = 6, < 0.05) than in a control. This indicated modifications of calcium handling by Epac that we examined next.

Figure 5.

 Modulation of rhythmic activity by calcium influx and release and the effects of Epac agonist. Calcium increases were induced by membrane depolarisation with 45 mM K+ for 10 s (a), 1 mM ATPs (b) and 10 mM caffeine (c) in a control and 3 min after addition of 0.3 μM 8-pCPT to the bath. Neuronal and single channel activities were enhanced and then depressed in parallel with long-lasting calcium elevations.

Calcium can be taken up and released from endoplasmatic reticulum (ER), the main internal storage site in different cell types (Berridge 2002). ER is functionally separated into the two pools which are sensitive to IP3 and caffeine, respectively. We examined the first mechanism using ATP to activate P2Y receptors and IP3-driven calcium release (Mironov 1994). The effects of 1 mM ATP were similar to those of depolarisation (Fig. 5b) and they were not observed in calcium-free solutions (= 3) that excludes possible involvement of ionotropic P2X receptors. Another IP3-related agonist, bradykinin (100 nM) evoked similar effects (= 6, data not shown). It did not increase calcium when applied directly after ATP (and vice versa) indicating depletion of IP3-sensitive pool after single application of IP3-producing agonist (Mironov 1994). In the presence of 8-pCPT, the responses to ATP became bigger (0.24 ± 0.03 vs. 0.16 ± 0.03 μM in a control, = 6, < 0.05) and rhythmic activity recovered faster (the right panel in Fig. 5b). Activation of calcium release from caffeine-sensitive stores showed marginal effects (Fig. 5c, n = 6). The responses were not modified in the presence of adenosine antagonist 8-Cyclopentyl-1,3-dipropylxanthine (DPCPX) (1 μM, = 3) or the agonist 2-Chloro-N6-cyclopentyladenosine (CCPA) (n = 3) that excludes contribution of adenosine receptors (Mironov et al. 1999) in the effects of caffeine.

We further examined the role of the two calcium pools in the effects of 8-pCPT using ryanodine which irreversibly activates caffeine-sensitive release channels and dantrolene which blocks them; by depleting ER with thapsigargin (SERCA blocker) and inhibiting phospholipase C (PLC) with U73122. Thapsigargin transiently elevated calcium that first potentiated TRPM4 channel and bursting activities which then subsided and were not restored with 8-pCPT (Fig. 6a, n = 4). Depletion of caffeine-sensitive pool with ryanodine only slightly depressed rhythmic activity which was then potentiated by 8-pCPT (Fig. 6b, n = 4).

Figure 6.

 Epac and internal calcium stores. (a) Transient enhancement of rhythmic calcium transients, bursting and single channel activities after depletion of internal stores with SERCA inhibitor thapsigargin. Subsequent application of 0.3 μM 8-pCPT did not restore repetitive activity. (b) Ryanodine weakly suppressed bursting but it was further enhanced by 8-pCPT. (c, d) Depletion of calcium stores with thapsigargin decreased the amplitude and duration of depolarisation-induced calcium transients to a greater extent than in a control (Fig. 5) and in the presence of ryanodine (d). (e) PLC blocker U73122 inhibited bursting and occluded the effects of 8-pCPT. (f) Dantrolene (a blocker of calcium release from caffeine-sensitive stores) weakly suppressed repetitive activity and did not modify its potentiation by 8-pCPT.

Brief depolarisations of neurons with depleted calcium stores produced tonic firing of shorter duration than in a control (Fig. 6c vs. Fig. 5a) that correlated with a shorter duration of calcium elevations. In the presence of thapsigargin, calcium increases decayed with the time-constant 6.8 ± 0.3 s that was distinctly smaller than in a control (= 4, < 0.01). This can be explained by the calcium-induced calcium release (Berridge 2002) that can augment calcium elevation because of calcium influx through voltage-sensitive calcium channels. It seemingly operates in a control but is absent when ER was depleted with thapsigargin. In the presence of ryanodine depolarisation-induced calcium elevations (0.30 ± 0.03 μM) and their decay (time-constant 12.6 ± 0.4 s) were not different from those in a control (Fig. 6d, n = 4, > 0.05).

After blockade of phospholipase C with U73122 the rhythmic activity disappeared and was not restituted by 8-pCPT (Fig. 6e). The inactive analogue U73343 was without effect (= 3). Blockade of calcium release from caffeine-sensitive pool with dantrolene slightly depressed bursting and 8-pCPT potentiated the activity above the control levels (Fig. 6f, n = 4). The frequency of bursts in a control, with dantrolene and 3 min after 0.3 μM 8-pCPT were 6.2 ± 0.4 per minute, 5.6 ± 0.5 per minute (> 0.05) and 9.4 ± 0.5 per minute (< 0.05), respectively, and the number of APs in the bursts in these experiments were 4.4 ± 1.1, 3.2 ± 1.0 (> 0.05) and 6.4 ± 1.2 (< 0.05). Taken together, the data indicate the importance of IP3-sensitive pool in bursting and the minor role of caffeine-sensitive calcium pool that is in line with previous findings in functional acute slices (Mironov 2008).

Inhibition of K-ATP channels by 8-pCPT

We next examined the role of Epac in cAMP-mediated suppression of K-ATP channels which modulate bursting in functional slice preparation (Mironov et al. 1999). In the organotypic slice, the channels had the same conductance and pharmacological properties (Fig. 7a) as in the inspiratory neurons in acute slices (Mironov et al. 1998). 0.3 μM 8-pCPT inhibited K-ATP channels that was accompanied by the enhancement of bursting activity (Fig. 7b, n = 6). Forskolin suppressed the activity of K-ATP channels and subsequent addition of 8-pCPT inhibited them further (Fig. 7c, n = 4). When K-ATP channels were inhibited by 8-pCPT, the potentiation by forskolin was not observed (= 4, data not shown).

Examination of possible downstream targets of Epac

Several mediators of Epac actions have been identified in the heart and muscle cells and we tested whether they are functionally present in neurons. For example, 8-pCPT modifies calcium release in cardiac myocytes through Calcium/calmodulin-dependent protein kinase II (Pereira et al. 2007) and Epac modulation of vascular K-ATP channels is attributed to activation of Ca2+-sensitive protein phosphatase 2B (Purves et al. 2009). After the treatments with calmodulin inhibitor calmidazolium (50 μM for 10 min) and cyclosporin A (20 μM for 30 min, used to block Protein phosphatase 2B) neither bursting activity and gating of TRPM4 and K-ATP channels, nor the effects of 8-pCPT were modified (= 3 for each experiment, data not shown).

We examined a possible role of PIP2, a precursor of IP3 which directly controls the activity of both TRPM4 (Nilius et al. 2007) and K-ATP channels (Kang et al. 2003, 2008). PIP2 pool can be depleted with wortmannin and enriched with bradykinin (Loew 2007; Pian et al. 2007). Incubation of slices with 1 μM wortmannin for 30 min increased mean burst frequency from 7.8 ± 0.2 to 12.2 ± 0.3 per minute (Fig. 8a, n = 8, < 0.05). Alone suppression of calcium release from ER which is IP3 (and PIP2) dependent can not explain the effects of wortmannin on TRPM4 channels. The maximal (minimal) popen values decreased nearly proportionally from 0.30 ± 0.02 (0.06 ± 0.02) in a control to 0.08 ± 0.01 (0.02 ± 0.01) after wortmannin treatment (= 4, < 0.01). The effects were significantly bigger than a decrease in the amplitude of calcium transients by 25 ± 4%. In the presence of bradykinin, pmax and pmin values were 0.68 ± 0.06 (0.32 ± 0.05) (= 4, < 0.05) and potentiation of activity was bigger than could be expected from the increase in the amplitude of calcium transients (from 0.23 ± 0.03 to 0.31 ± 0.01 μM, < 0.05). Differential effects on calcium fluctuations and channel activity can be explained by direct stimulation of TRPM4 channels by PIP2. In the presence of wortmannin and bradykinin, 8-pCPT induced proportional changes in the amplitude of calcium transients and pmax and pmin values (both by about 50%) that likely reflected sensitisation of calcium release by Epac.

Figure 8.

 Modulation of TRPM4 and K-ATP channels by wortmannin and bradykinin. (a) Suppression of TRPM4 channel activity after treatment with wortmannin (1 μM for 30 min) and their potentiation by bradykinin (100 nM for 3 min with following incubation for 30 min) used to modulate PIP2 levels. The treatments produced opposite changes in bursting and proportionally scaled the effects of 8-pCPT. (b) Modulation of activity of K-ATP channels after wortmannin and bradykinin treatments and the effects of 8-pCPT.

Wortmannin suppressed the activity of K-ATP channels and their pmax (pmin) values decreased from 0.33 ± 0.01 (0.16 ± 0.02) in a control to 0.16 ± 0.02 (0.08 ± 0.01) (= 4, < 0.01). After treatment with bradykinin they increased to 0.61 ± 0.04 (0.32 ± 0.03) (= 4, < 0.01, Fig. 8b). A proportional scaling of pmin and pmax in these experiments can be explained by direct PIP2–K-ATP channel interactions. Application of 8-pCPT after the treatments proportionally decreased pmax and pmin values (Fig. 8b) that likely indicates independent actions of Epac and PIP2 on K-ATP channels.

8-pCPT effects in functionally intact preparation

In the last set of experiments, we tested the effects of Epac in the functionally intact slices which were isolated as described in Methods. In six slices examined 0.3 μM 8-pCPT potentiated the bursts of inspiratory activity, TRPM4 channels and the integrated motor output (Fig. 9a). The effects were identical to those observed in the organotypic slices (Fig. 2). After wash-out of 8-pCPT for 30 min the variables returned to the initial levels and we then treated the slices with 1 μM wortmannin for 30 min or 100 nM bradykinin (3 min application, 30 min wash-out). The drugs showed opposite changes in rhythmic activity which closely resembled those observed in organotypic brainstem slices (Fig. 8). Subsequent application of 8-pCPT also potentiated repetitive activity and the effects were similar to those observed in organotypic slices (Fig. 9b and c).

Figure 9.

 Epac effects in preBötC neurons in functionally intact slices. Recordings were made in preBötC neurons in the functional brainstem slices isolated according to Smith et al. (1991) as described previously (Mironov et al. 1998). DIC image is shown in top inset. The effects of 8-pCPT application were examined in a control and after treatments with 1 μM wortmannin and 100 nM bradykinin (3 min, washout with ACSF for 30 min). The two traces in each recording show TRPM4 channel and bursting activities in the inspiratory neurons and integrated motor output (∫XII). The graphs on the right present changes in mean interburst interval (empty circles) and the number of APs within the bursts (filled circles) during the applications.


In this study, we examined the actions of novel cAMP effector, Epac, in preBötC, an essential part of the respiratory network (Feldman and Del Negro 2006). The main contributions of the present work are as follows: Epac (i) potentiates IP3-driven calcium release from internal stores; (ii) stimulates TRPM4 channels, (iii) inhibits K-ATP channels and (iv) all three effects contribute to the reinforcement of rhythmic bursting activity in organotypic and functionally intact slices. (v) TRPM4 and K-ATP channel activities are modified by treatments directed to modulate PIP2 levels that also modulates bursting activity. The novel pathways examined in this study may regulate the function of preBötC neurons and be involved in reinforcement of breathing activity by cAMP.

cAMP-signalling systems provide a pivotal centre for achieving cross-talk regulation by various signalling pathways. Stimulatory and inhibitory neurotransmitters often modulate neuronal activities through changes in cAMP that modifies specific ion conductances (Rekling et al. 2000). cAMP elevation produces a long-term stabilisation of rhythmic motor output (Mironov et al. 1999; Ruangkittisakul et al. 2006) that could involve activation of PKA able to phosphorylate L-type calcium (Mironov and Richter 1998) and glutamate-gated channels (Shao et al. 2003). In this study, we examined the role of Epac, a novel effector of cAMP, which selectively controls many important physiological processes in the heart, pancreas and multiple cell lines (Bos 2003; Seino and Shibasaki 2005; Holz et al. 2006). Little is known about the actions of Epac in CNS but several recent studies show that 8-pCPT modulates synaptic transmission (Zhong and Zucker 2005) and plasticity (Ster, 2009; Ma et al. 2009), the effects which have been previously assigned to PKA (Siegelbaum and Kandel 1991). The knock-out animals are not yet available (Holz et al. 2006) and therefore the experimental strategy by which the existence of Epac signalling pathway is validated relies on the use of specific Epac activator, 8-pCPT.

We used organotypic brainstem slices (Hartelt et al. 2008) prepared to contain all specific nuclei (Ruangkittisakul et al. 2006) which serve as the hallmarks to define preBötC in the transverse brainstem sections. The preparation has both advantages and disadvantages. In order to maintain the viability of slices in culture, they have to be made thinner (250 μm) than acutely isolated and functionally intact slices (∼ 600 μm). Although not all neurons from preBötC were isolated, a functional calcium imaging revealed periodic activity (Fig. 1) that closely resembled a functional motor output (Fig. 9). Current ideas about rhythm generation within preBötC (Feldman and Del Negro 2006; Mironov 2009) imply that preBötC neurons, which default mode may be quiescent, are dynamically engaged in producing emergent synchronous activity via glutamatergic synaptic transmission. This pattern can be generated by small ensembles of neurons (Hartelt et al. 2008) whose synchronous discharges are coordinated by calcium waves propagating along neuronal processes (Mironov 2008). The mechanism reminesces the one proposed for integration of neuronal activity in neocortex (Loewenstein and Sompolinsky 2003). Substantial redundancy of the network elements in preBötC may justify their examination in isolation in simple preparations such as organotypic slices. The neurons endow the same channel types as their in vivo counterparts and express Neurokinin 1-receptors (Hartelt et al. 2008) thought to represent a hallmark of respiratory neurons (Gray et al. 2001). The neurons demonstrate intrinsic bursting activity which pattern is similar to that of inspiratory neurons in the functionally intact slices (Smith et al. 1991; Mironov et al. 1998, 1999; Mironov 2008). The bursting is also accompanied by repetitive calcium transients (Mironov et al. 2009a), a readout of rhythmic activity generated by preBötC network in acute slices (Koshiya and Smith 1999; Mironov 2008). Rhythmic activity in organotypic slices disappeared after blockade of AMPA receptors with CNQX (Fig. 1c) reproducing a well-known effect in the functional acute slices (Koshiya and Smith 1999; Mironov 2008). The slice culture preparation is well suited for selective transduction of neurons that we have used to map the topology of neuronal network within preBötC (Hartelt et al. 2008) and for functional imaging of calcium and cAMP (Mironov et al. 2009a,b).

Epac agonist 8-pCPT potentiated calcium release from IP3-sensitive stores that stimulated TRPM4 channels responsible for depolarisation drive and generation of bursts (Mironov 2008). 8-pCPT also suppressed the activity of K-ATP channels (Fig. 7). They are abundantly expressed in the respiratory neurons (Mironov et al. 1998), contribute to the generation of the inspiratory activity (Pierrefiche et al. 1996; Haller et al. 2001), are modulated by cAMP (Mironov et al. 1999) and determine the respiratory response to oxygen deprivation (Mironov et al. 1998). Their inhibition by Epac reveals a new facet of K-ATP channel modulation in neurons that may have particular physiological significance by coupling of metabolic events to neuronal activity (MacGregor et al. 2002).

Forskolin stimulates cAMP synthesis through activation of adenylate cyclase. Potentiation of bursting activity by forskolin was independent from PKA and was reinforced by 8-pCPT. The results can be explained on the basis of previous cAMP measurements (Mironov et al. 2009b). 1 μM forskolin increases cAMP in preBötC neurons from about 0.5 to 1.5 μM that (according to EC50 = 1.5 μM cAMP) would increase Epac activity from 25% to 50%. On the other hand, 0.3 μM 8-pCPT (EC50 = 40 nM) should activate Epac to 90% i.e. much stronger than forskolin does that abolishes its effects.

The effects of Epac stimulation examined in this study are contingent with a scheme (Fig. 10) that includes the elements described previously in other cell types (Bos 2003; Seino and Shibasaki 2005; Holz et al. 2006). We propose that sensitization of calcium release from IP3-sensitive stores by Epac stimulates TRPM4 channels. K-ATP channels are inhibited by Epac that additionally potentiates bursting. TRPM4 and K-ATP channels can adjust their activity in relation to PIP2 levels as determined by dynamical equilibrium established by PLC and phosphoinositide 3-kinases/Phosphatases.

Figure 10.

 Epac-related signalling pathways examined in this study. Activation of P2Y receptors by ATP stimulates Gq and phospholipase C (PLC) producing IP3 from PIP2. Forskolin activates adenylate cyclase (AC) that leads to cAMP-dependent stimulation of Epac which inhibits K-ATP channels and sensitizes IP3 receptors. Calcium release from IP3-stores potentiates the activity of TRPM4 channels. Caffeine-sensitive ryanodine receptors (RyR) are not modulated by Epac. TRPM4 and K-ATP channels are regulated by PIP2 which levels can be increased after activation of bradykinin receptors. Blockade of PI3 and PI4 kinases with wortmannin depletes PIP2 supplies because of persistent activity of phosphatases (PP).

Phosphatidylinositol 4,5-bisphosphate actions on ion channels represent a novel and important mechanism of regulation of their activity (see Suh and Hille 2008 for recent review). PIP2 levels are controlled by PLC and Epac in the heart (Oestreich et al. 2009) and receptor-mediated activation of PLC depletes PIP2 supplies that can lead to inhibition of K-ATP channels through the lowering of the affinity to ATP (Baukrowitz et al. 1998; Shyng and Nichols 1998). We used bradykinin and wortmannin to modulate PIP2 pool (Loew 2007; Pian et al. 2007). Bradykinin stimulated TRPM4 and K-ATP channels and after wortmannin their activities decreased. The effects would be in line with presumed enrichment or depletion of PIP2 levels, yet the data are indirect. More insight about PIP2 interactions with TRPM4 and K-ATP channels can be gained with new probes for PIP2 imaging (Wuttke et al. 2010) and examination of inside-out patches where non-hydrolyzable PIP2 analogues potentiate TRPM4 channels (Mironov 2008).

Phosphatidylinositol 4,5-bisphosphate-related effects deserve attention in considering the mechanisms of adaptation of persistent activity in neuronal networks to different environmental conditions that can be especially important for breathing. PIP2-channel interactions may alter neuronal activity for much longer periods of time than calcium and cAMP do which actions comprise the time scales of ∼ 10 s and ∼ 1 min, respectively (Mironov et al. 2009a,b). Crowder et al. (2007) proposed the active role of PIP2 in counteracting desensitization of TRPM4 channels that enhances the robustness of respiratory rhythm. Our data are in line with this suggestion and we additionally propose PIP2 effects on K-ATP channels. PIP2 is enriched in dendritic branching points (Jacob et al. 2005) where calcium waves are often initiated (Larkum et al. 2003; Mironov 2008). Concerted actions of Epac and PIP2 on calcium release, TRPM4 and K-ATP ion channels can provide a flexible and long-lasting functional feedback regulating cAMP-dependent effects on breathing. PIP2 actions may be not restricted to post-synaptic site (examined in the study) but could be spread out to the pre-synaptic ending such as in hippocampal neurons where pre-synaptic accumulation of PIP2 is induced by post-synaptic activity via a retrograde action of nitric oxide (Micheva et al. 2001). Combination of pre- and post-synaptic events is crucial in long-term potentiation induction (Siegelbaum and Kandel 1991) and their interplay may be important in strengthening of synapses and consolidation of activities in neuronal networks including preBötC.


We greatly appreciate the gift of D3cpv-camps construct from S. Kügler and are indebted to Nicole Hartelt for expert technical assistance and Cathy Ludwig for thorough reading of the manuscript. The work was supported by DFG grant MI 685/2-1 and funded by Deutsche Forschungsgemeinschaft through the DFG-Research Center for Molecular Physiology of the Brain.