We have previously used cyclic nucleotide-gated (CNG) channels as sensors to measure cAMP signals in human embryonic kidney (HEK)-293 cells. We found that prostaglandin E1 (PGE1) triggered transient increases in cAMP concentration near the plasma membrane, whereas total cAMP levels rose to a steady plateau over the same time course. In addition, we presented evidence that the decline in the near-membrane cAMP levels was due primarily to a PGE1-induced stimulation of phosphodiesterase (PDE) activity, and that the differences between near-membrane and total cAMP levels were largely due to diffusional barriers and differential PDE activity. Here, we examine the mechanisms regulating transient, near-membrane cAMP signals. We observed that 5-min stimulation of HEK-293 cells with prostaglandins triggered a two- to threefold increase in PDE4 activity. Extracellular application of H89 (a PKA inhibitor) inhibited stimulation of PDE4 activity. Similarly, when we used CNG channels to monitor cAMP signals we found that both extracellular and intracellular (via the whole-cell patch pipette) application of H89, or the highly selective PKA inhibitor, PKI, prevented the decline in prostaglandin-induced responses. Following pretreatment with rolipram (a PDE4 inhibitor), H89 had little or no effect on near-membrane or total cAMP levels. Furthermore, disrupting the subcellular localization of PKA with the A-kinase anchoring protein (AKAP) disruptor Ht31 prevented the decline in the transient response. Based on these data we developed a plausible kinetic model that describes prostaglandin-induced cAMP signals. This model has allowed us to quantitatively demonstrate the importance of PKA-mediated stimulation of PDE4 activity in shaping near-membrane cAMP signals.
- G protein signaling
- protein kinase A
- A-kinase anchoring protein
- CNG channel
cyclic amp (cAMP) is an ubiquitous second messenger known to regulate diverse cellular functions (45). Although the basic mechanisms by which cAMP signals transmit information to downstream effectors have been well established, we are only beginning to unravel how these signals differentially regulate a wide variety of cellular responses over vastly different timescales. One hypothesis is that cAMP signals are compartmentalized in different regions of the cell. In fact, this hypothesis emerged over 20 years ago to explain how different extracellular stimuli that primarily act through cAMP can have distinct downstream effects on the cell (4, 44). In recent years several groups have presented evidence implicating diffusional restrictions (31, 32), phosphodiesterase (PDE) activity (1, 23, 26, 32, 34, 49), and buffering by protein kinase A (PKA) (2, 7, 33, 39) as critical factors in shaping cAMP signals. Given the importance of these signals in cellular physiology, it is striking that little is known about the kinetics and subcellular localization of cAMP. One reason for the dearth of information is that until recently there have been no practical, single-cell cAMP sensors. In the past several years a series of genetically-encoded cAMP sensors based on cyclic nucleotide-gated (CNG) channels (31, 34), PKA (26), and EPAC (exchange proteins directly activated by cAMP) (8, 27, 29) have been developed. Direct binding of cAMP to each of these proteins triggers a conformational change that can be monitored by fluorescence resonance energy transfer (EPAC and PKA) or electrophysiological methods (CNG channels). The time course of these conformational changes has been best studied for CNG channels and PKA. The overall kinetics of CNG channels are much faster than PKA, and CNG channels have a much lower buffering capacity than heterologously expressed PKA. These factors make CNG channels better suited to measure fast cAMP responses near the plasma membrane (reviewed in Ref. 33). Less is known about the kinetics and buffering capacity of the EPAC-based sensors, but they do have potential advantages over PKA-based sensors (lower cAMP buffering capacity), and CNG channels (monitoring cytosolic cAMP signals) (see Refs. 8, 27).
When using these approaches one must be keenly aware of their advantages and disadvantages, and pick the sensor best suited to measure a specific signal. Two critical factors are the sensor's kinetics and buffering capacity (33). We have previously taken advantage of the rapid kinetics and low buffering capacity of CNG channel-based sensors to examine the subcellular compartmentalization of cAMP signals. We observed that 1) cAMP concentrations near the plasma membrane were >10-fold higher than total cAMP levels in C6–2B glioma cells and human embryonic kidney (HEK)-293 cells stably overexpressing adenylyl cyclase (AC) type 8; 2) cAMP accumulation at the plasma membrane was resistant to dialysis in the whole-cell, patch-clamp configuration; and 3) the wash-in of cAMP from the patch pipette to the channels was anomalously slow (31). In addition, we developed a compartmental model of cAMP signaling that was able to describe all of these results. The model predicted that in response to a stimulus, kinetically distinct cAMP signals could be observed within cells, and that cyclic nucleotides produced from distinct pools of cyclase (e.g., particulate and soluble guanylyl cyclase) would not have the same efficacy in activating CNG channels, even if similar total cellular cyclic nucleotide levels were reached. We have subsequently provided evidence supporting both predictions (28, 32). As part of the former study we found that prostaglandin E1 (PGE1) triggers a transient cAMP signal near the plasma membrane of these cells, and presented evidence that the decline in the signal was primarily due to a PGE1-induced stimulation of PDE activity.
Here, we have used CNG channel-based sensors to more closely examine the underlying mechanisms responsible for transient cAMP signals near the plasma membrane. Our results demonstrate that PKA-mediated stimulation of PDE4 activity underlies the decline in transient cAMP signals, and that the subcellular localization of PKA via A kinase anchoring proteins (AKAPs) is critical in the regulation of these signals. We used these data to develop a plausible kinetic model of near-membrane cAMP signals. The simulations demonstrate that stimulation of PDE4 activity is sufficient to cause the observed decline in transient cAMP signals. Furthermore, based on these simulations we propose that high local concentrations of PKA serve to buffer cAMP, contributing to its slow spatial spread within cells (23, 28, 31, 40). Resolving the kinetics of the near-membrane cAMP signals in this relatively simple cellular system should help us better understand the physiological consequences of PDE4 regulation and the mechanisms by which cAMP is localized to different cellular compartments.
MATERIALS AND METHODS
Cell culture and channel expression.
HEK-293 cells were maintained in culture and transfected using adenovirus constructs as described previously (31). Briefly, HEK-293 cells were maintained in 10 ml of MEM (Life Technologies) with 10% (vol/vol) fetal bovine serum (Gemini), and grown in 100-mm culture dishes at 37°C in a humidified atmosphere of 95% air-5% CO2. Cells were plated at ∼60% confluence in 100-mm dishes for infection with the C460W/E583M CNG-channel-encoding adenovirus construct (multiplicity of infection ∼10 plaque-forming units/cell) (34). Two hours postinfection, hydroxyurea was added to the cell media at 1 mM final concentration to inhibit viral replication. Twenty-four hours postinfection cells were detached with phosphate-buffered saline containing 0.03% EDTA, resuspended in serum-containing media, and assayed within 12 h. All experiments were conducted at room temperature, 20–22°C. Unless otherwise stated all reagents were purchased from Sigma.
Measurement of PDE activity.
Cyclic AMP PDE activity was measured according to the method of Thompson and Appleman (46) as detailed previously (16). Briefly, after incubation for 1–5 min with or without PGE2 (1 μM), forskolin (50 μM), or isoproteronol (10 μM), cells were harvested and homogenized in ice-cold hypotonic buffer (in mM), 20 Tris·HCl, pH 8.0, 50 NaF, 1 EDTA, 0.2 EGTA, 10 Na2PO4, 5 β-mercaptoethanol, and a protease inhibitor cocktail [leupeptin, 0.5 μg/ml; aprotinin, 4 μg/ml; benzamidine, 50 mM; pepstatin, 0.7 μg/ml; soybean trypsin inhibitor, 10 μg/ml; and PMSF, 10 μg/ml (freshly added before use)] using an all-glass homogenizer. In some experiments, homogenates were centrifuged at 14,000 g for 15 min and the supernatant (soluble) and pellets (particulate fraction) were assayed separately. Aliquots of the homogenates or of the fractionated extracts were assayed for PDE activities with 1 μM [3H]cAMP as a substrate. PDE4 activity was defined as the fraction of cAMP PDE activity inhibited by 10 μM rolipram (a PDE4 inhibitor). Protein concentrations were determined using the Bio-Rad protein assay (Bio-Rad Laboratories, Hercules, CA) with BSA as a standard. Experiments were repeated at least three times.
Western blot analysis.
After incubation with the indicated substances, cells were homogenized in ice-cold hypotonic buffer and the protein concentration was determined. Samples (40 μg protein) were boiled in Laemmli buffer (25), subjected to electrophoresis on an 8% SDS-PAGE gel, and blotted onto Immobilon-P transfer membrane (Millipore, Bedford, MA). Membranes were blocked in TBS-0.1% Tween 20 containing 5% nonfat milk. The phosphorylated and total cAMP response element binding (CREB)/activating transcription factor (ATF) proteins were detected using mouse monoclonal (Upstate Biotechnology) and rabbit polyclonal antibodies (Cell Signaling), respectively and visualized by use of ECL detection reagents (Amersham Pharmacia Biotech).
Measurement of total cellular cAMP levels.
HEK-293 cells were plated at 33% confluence in 12-well plates and assayed 24–48 h later. Cells were washed and assayed in a solution containing (in mM), 145 NaCl, 4 KCl, 10 HEPES, 10 d-glucose, 1 MgCl2, 1 CaCl2, pH 7.4. Additions were made from 100× stock solutions. Reactions were terminated by addition of 1 N HCl (0.1 N HCl final) and plates were incubated on ice for 15 min, after which the cells were scraped from the well. Cellular cAMP levels were measured using enzyme immunoassays (Direct Cyclic AMP Enzyme Immunoassay Kit, Assay Designs). Sample cAMP concentrations were calculated from standard curves. Data are presented as means ± SE, performed in triplicate.
Monitoring near-membrane cAMP signals in cell populations.
Cyclic AMP signals were monitored in cell populations as described previously (32–34). Briefly, we took advantage of the Ca2+ permeability of CNG channels comprised of the rat olfactory channel α subunits, CNGA2 (14), and measured the rate of Ca2+ influx to monitor changes in cAMP levels. In this assay, an increase in local cAMP concentration causes activation of CNG channels and a subsequent increase in the rate of Ca2+ entry (12, 31–34). Changes in the rate of Ca2+ influx in response to stimuli reflect changes in the cAMP levels. We used the fluorescent indicator fura-2 to monitor Ca2+ influx in cell populations. Cells were loaded with 4 μM fura-2 AM (the membrane-permeant form, Calbiochem) at room temperature, 20–22°C, for 30–40 min, in MEM supplemented with 20 mM HEPES, pH 7.4. Cells were washed twice, then resuspended in a solution containing (in mM), 145 NaCl, 11 d-glucose, 10 HEPES, 4 KCl, 1 CaCl2, and 1 MgCl2, pH 7.4 (3–4 × 106 cells/3 ml buffer solution), and assayed using a PTI DeltaScan-1 spectrofluorimeter (Photon Technology International). PGE1, PGE2, rolipram, and the PKA inhibitor H89 (Calbiochem) were added to a stirred cuvette from concentrated DMSO stocks, with final concentrations as indicated (final DMSO concentrations were ≤0.2%). The mixing time was estimated to be on the order of 5 s. Fluorescence was measured at an excitation wavelength of 380 nm and an emission wavelength of 510 nm. Under these conditions Ca2+ influx caused a decrease in fluorescence (ΔF), which was expressed relative to the prestimulus fluorescence (F0) to correct for variations in dye concentration, and to allow for comparison of results on different batches of cells. Data were sampled at either 5 or 10 Hz and filtered at 0.5 or 1 Hz. All data are representative of at least four experiments.
To ensure that extracellular application of H89 did not alter the Ca2+ handling properties of HEK-293 cells, we examined the effects of a 10 min pretreatment with 10 μM H89 on thapsigargin-induced changes in intracellular Ca2+ levels. Pretreatment with H89 did not significantly change the Ca2+ response induced by 1 μM thapsigargin in nominally Ca2+-free conditions (Ca2+ release from internal stores), or in the presence of 1 mM extracellular Ca2+ (Ca2+ release from internal stores and subsequent Ca2+ entry). Thus, it is unlikely that H89 significantly altered the Ca2+ handling properties of the cells.
Monitoring near-membrane cAMP signals in single cells.
Single-cell cAMP measurements were made using the whole-cell, patch-clamp technique. Recordings were made using an HEKA EPC10 patch-clamp amplifier. To ensure adequate voltage control, pipette resistance was limited to 4 MΩ and averaged 2.6 ± 0.2 MΩ (n = 43). Voltage offsets were zeroed with the pipette in the bath solution; no additional corrections were made for the liquid junction potential difference. Experiments with a series resistance-induced error in excess of 5 mV were discarded. After achieving whole-cell configuration, the preparation was allowed to equilibrate for at least 10 min to ensure sufficient time for dialysis of compounds from the patch pipette into the cell. Current records were typically sampled at 10 kHz and filtered at 2 kHz and stored on a PC. Currents were recorded during 400-ms steps to a membrane potential of +20 mV from a holding potential of 0 mV. The pipette solution contained (in mM) 140 KCl, 0.5 MgCl2, 10 HEPES, 5 Na2ATP, 0.5 Na2GTP, pH 7.4; the bath solution contained (in mM) 140 NaCl, 4 KCl, 10 d-glucose, 10 HEPES, and either 0.1 or 10 MgCl2, pH 7.4. PGE1 and rolipram (Calbiochem) were added to control solutions from concentrated DMSO stocks (final DMSO concentrations ≤0.2%), with final concentrations as indicated. PKI (Calbiochem), stearated (St)-Ht31 (Promega), H89, and St-Ht31P control peptide (Promega) were aliquoted as 1,000× stock solutions and stored at −20°C. Solutions were applied using the SF-77B fast-step solution switcher (Warner Instruments). The mechanical switch time was 1–2 ms. The time required to exchange the extracellular solution was measured by applying a 140 mM KCl solution to a depolarized cell (+50 mV) and monitoring changes in current through endogenous voltage-gated K+ channels; for each experiment, it was less than 100 ms. The bulk solution within the bath chamber was changed within 20 s using a custom-built, gravity-driven perfusion system.
Data analysis and mathematical simulations.
All data were analyzed using custom scripts written in the MATLAB programming environment (MathWorks) and statistical analysis was performed using SigmaPlot (v.9; Systat Software) using Student's t-test. Electrophysiological data were converted to formats compatible with MATLAB software using a custom script provided by Bruxton. Simulations were performed using the fourth-order Runge-Kutta solver in the MATLAB programming environment.
Recent studies have demonstrated that PDE4D5 and PDE4D9 are responsible for the majority of endogenous PDE activity in HEK-293 cells (36). Consistent with these results, we have demonstrated that PDE4 is primarily responsible for the IBMX-sensitive PDE activity near the plasma membrane of these cells (34). We have also provided evidence suggesting that stimulation of PDE4 activity was largely responsible for the decline in cAMP levels following PGE1-stimulation of HEK-293 cells (32). Here we sought to determine the mechanisms that underlie feedback regulation of cAMP signals near the plasma membrane of these cells. One interesting possibility is that PKA-mediated phosphorylation of PDE4 increases PDE activity, causing the observed decline in the transient cAMP signals.
Effects of H89 on PDE4 and AC activity.
To determine the extent to which prostaglandins induce activation of PDE4, we exposed HEK-293 cells to vehicle or 1 μM PGE2 for different times and then measured PDE4 activity in either the crude homogenate, or in the soluble and particulate fractions. PDE4 activity was determined as the rolipram-sensitive PDE activity. PGE2 treatment triggered a two- to threefold increase in PDE4 activity in both the soluble and particulate fractions of the cells (Fig. 1, A and B) as well as in the homogenate (data not shown). An increase in activity was detected as early as 1 min after PGE2 addition (Fig. 1, A and B). The increase in PDE4 activity was recovered in the pellet after immunoprecipitation with PDE4D-selective antibodies (data not shown), suggesting that activation of this isoenzyme plays a major role in this G protein-coupled receptor (GPCR)-dependent regulation of cAMP signals. Pretreatment of cells for 10 min with the PKA inhibitor H89 reduced the PGE2-induced stimulation of PDE4 activity in a dose-dependent manner (Fig. 1C). Half-maximal inhibition of PGE2-stimulated PDE activity occurred following pretreatment with ∼4 μM H89, and complete inhibition occurred at 10 μM H89. Similar results were observed following 10 μM isoproterenol-induced or 100 μM forskolin (an AC activator)-induced stimulation of cAMP production (data not shown).
The extracellular concentrations of H89 used in these experiments were significantly higher than the reported KI for PKA, ∼40 nM (17), and could potentially inhibit other protein kinases. However, under these conditions we were unable to estimate the intracellular H89 concentrations. To ensure that H89 was indeed inhibiting PKA, we tested the dose dependence of H89 inhibition of CREB phosphorylation in the same preparation. Under these experimental conditions, H89 inhibited CREB phosphorylation with a similar dose dependence as it inhibited PDE4 stimulation (Fig. 1D). In control experiments, pretreatment with 1–30 μM H85, an analog of H89 that does not inhibit PKA, had little or no effect on prostaglandin-induced stimulation of PDE4 activity or CREB phosphorylation (data not shown).
We next examined the effects of H89 and rolipram on forskolin-, PGE1-, and PGE2-induced total cellular cAMP accumulation using enzyme immunoassays (Fig. 2). In response to vehicle alone (5 min) 0.9 ± 0.3 pmol/well of cAMP was measured, and there was no significant increase in cAMP levels following 10 min pretreatment with 10 μM H89 alone (open bars), pretreatment with 10 μM rolipram alone (cross-hatched bars) or H89 and rolipram (hatched bars). Stimulation with a subsaturating concentration of forskolin (1 μM, 5 min) triggered no significant increases in total intracellular cAMP, whereas PGE1 and PGE2 (1 μM, 5 min) triggered significant increases (∼3-fold over vehicle alone). Pretreatment with H89, rolipram, or H89 + rolipram triggered significant increases in forskolin- and prostaglandin-induced cAMP accumulation. There were no significant differences between the prostaglandin-induced cAMP accumulation observed following pretreatment with rolipram alone vs. H89 + rolipram, indicating that the primary effect of H89 was the inhibition of prostaglandin-induced stimulation of PDE4 activity. This is consistent with our previous work that demonstrated the rate of total cAMP accumulation is nearly linear in the presence of PDE inhibitors (see figure 1B of Ref. 32).
Effects of H89 on cAMP signals monitored in cell populations.
We next sought to determine the effects of H89 on the prostaglandin-induced cAMP transients that occur near the plasma membrane of HEK-293 cells. We used a previously developed CNGA2 ion channel construct, containing the mutations C460W and E583M (C460W/E583M channels), to measure near-membrane cAMP signals. These mutations make it possible to measure cAMP concentrations in the 100 nM range, while rendering the channel relatively insensitive to cGMP (32, 34). C460W/E583M channels were heterologously expressed using a previously described adenovirus construct (34). To determine the effects of extracellular application of H89 on near-membrane cAMP signals we monitored changes in CNG channel activity by measuring Ca2+-influx through the channels; the fluorescent indicator fura-2 was used to monitor changes in intracellular Ca2+ levels in cell populations. With this approach, incremental changes in cAMP concentration are readily detected as changes in relative Ca2+ influx rates through C460W/E583M channels. By observing CNG channel responses in cell populations we avoid the possibility of inadvertently biasing our data by selecting single cells that are best suited for electrophysiological measurements. As such, we consider these important control experiments that allow us to ensure data from single cells are consistent with data from cell populations.
Figure 3A shows a typical response to 0.1 and 1 μM PGE1 in populations of HEK-293 cells (3–4 × 106 cells per cuvette) expressing C460W/E583M channels. Addition of PGE1 (added at 1 min, arrow) triggered an initial influx of Ca2+ through C460W/E583M channels, followed by a slower reduction in Ca2+ levels. Little or no response was observed in cells not expressing C460W/E583M channels. Pretreatment with 10 μM rolipram (3 min), a selective PDE4 inhibitor, prevented the decline in the responses triggered by either 0.1 or 1 μM PGE1 (Fig. 3B). Ten minute pretreatment with 10 μM H89 dramatically reduced the decline in the PGE1-induced responses (Fig. 3C). We determined that slopes of the PGE1-induced responses following 10 min pretreatment with 10 μM H89 and 3 min pretreatment with 10 μM rolipram were not significantly different than slopes of the PGE1-induced responses following pretreatment with rolipram alone (Fig. 3D). In control experiments, pretreatment with 10 μM H85 had little or no effect on prostaglandin-induced cAMP transients (data not shown). Also, pretreatment with 10 μM H89 had little or no effect on the time course or amplitude of Ca2+ influx in response to extracellular application of 100 μM pCPT-cGMP (a membrane permeant cGMP analog, data not shown), indicating that pretreatment with 10 μM H89 had little or no effect on CNG channel activity. These data are consistent with the hypothesis that PKA-mediated stimulation of PDE4 activity is primarily responsible for the decay in the cAMP response.
Effects of disrupting PKA-mediated signaling on cAMP signals measured in single cells.
Thus far we have focused our studies on using H89 to inhibit PKA-mediated effects in cell populations. By examining the dose-dependence of H89-mediated inhibition of CREB phosphorylation, we were able to choose an appropriate H89 concentration to inhibit PKA activity. In addition, we used H85 to control for potential non-kinase-specific effects of H89. However, it is possible that a 10 min pretreatment with 10 μM H89 may inhibit other kinases. To directly address this limitation we examined the effects of H89, PKI, a specific PKA inhibitor, and rolipram on cAMP signals measured in single cells using the whole-cell, patch-clamp technique. This high-resolution approach allows us to accurately measure the kinetics of prostaglandin-induced cAMP signals in single cells. In addition, compounds such as H89 or PKI can be introduced into the cell at known concentrations via the whole-cell patch pipette. Thus, we can accurately examine the dose-dependence of H89 or PKI effects on prostaglandin-induced cAMP signals. All electrophysiological experiments were conducted with nominally Ca2+-free buffers. Under these conditions the primary charge carriers through CNG channels are Na+ and K+. Thus, Ca2+ influx through CNG channels (and subsequent Ca2+-CaM mediated inhibition of CNG channel activity) did not contribute to the observed responses. No PGE1-induced responses were observed in control cells (cells not expressing CNG channels) or in cells expressing wild-type CNGA2 subunits (K1/2 for cAMP ∼40 μM).
Figure 4A shows a typical response of an HEK-293 cell expressing C460W/E583M channels to 1 μM PGE1. In this cell, PGE1 triggered a transient increase in outward current through C460W/E583M channels. This response is consistent with the PGE1-induced responses that we previously reported (32). Pretreatment of cells with 10 μM rolipram largely prevented the decline in the response (Fig. 4, B and D). Addition of rolipram following a PGE1-induced transient response triggered a large increase in current through CNG channels (Fig. 4C). These data demonstrate that the exposure of cells to PGE1 does not trigger cAMP accumulation to levels that saturate CNG channels, and are consistent with calibrated measurements of cAMP concentration near CNG channels, 0.7 ± 0.4 μM (32). To assess the extent to which rolipram prevented the decline in the cAMP response, we measured the percentage current remaining 3 min after peak current. In cells in which vehicle (0.1% DMSO) was added to the patch pipette solution 26 ± 6% of the peak current remained, whereas in cells that were pretreated with 10 μM rolipram, 92 ± 6% of the peak current remained (Fig. 4D).
We examined the effects of known intracellular H89 concentrations on the cAMP signals monitored in single cells (Fig. 5). In the presence of 100 nM H89 (∼2.5-fold higher than the KI for PKA, and significantly lower than the KI for other kinases), the rise in PGE1-induced responses was similar to that observed in control cells, whereas the decay was markedly less pronounced (compare Fig. 5A with Fig. 4A). At 200 nM H89 the decay was even less pronounced (Fig. 5B). Addition of H89 to the patch pipette increased the percentage of current remaining in a dose-dependent manner (a maximum of ∼68% in the presence of 0.5 or 1 μM H89, Fig. 5C). To further demonstrate the role of PKA-mediated feedback control of cAMP signals we used the highly specific PKA inhibitor PKI, KI ∼ 2.3 nM (6). We introduced PKI into the cell via the patch pipette because it does not readily cross the plasma membrane. In the presence of either 5 or 20 nM PKI, the decay of the responses was dramatically blunted (Fig. 6, A and B), with 81 ± 6% and 85 ± 4% current, respectively, remaining 3 min after the peak response (Fig. 6C). Taken together, the results presented thus far strongly indicate that PKA-mediated stimulation of PDE4 activity is required for the decline in the PGE1-triggered transient cAMP response.
We next sought to determine whether disruption of AKAPs would also alter the kinetics of PGE1-stimulated cAMP signals. We monitored PGE1-induced responses with an AKAP inhibitor peptide, Ht31, or the associated control peptide, Ht31P, in the patch pipette (Fig. 7). Ht31 is a peptide with an amphipathic helix motif that is a competitive inhibitor of all known PKA-AKAP interactions. Ht31P is a related peptide that has a proline in the sequence, which disrupts the amphipathic helix and binding to AKAPs (5). We found that including 10 μM Ht31 in the patch pipette solution caused a delay in and reduction of the rate of decay of PGE1-induced cAMP signals (Fig. 7A). In fact, 93 ± 3% of the current remained 3 min after the peak response (Fig. 7C). Transient PGE1-induced responses were still evident when 10 μM control peptide was included in the patch pipette (Fig. 7, B and C). These data clearly show the profound effects that the disruption of AKAPs has on cAMP signals.
Roles of PKA-mediated stimulation of PDE4 and cAMP buffering in shaping cAMP signals.
To further investigate the relative contributions of PDE activity and buffering in shaping cAMP signals we developed a plausible kinetic model of cAMP turnover near the plasma membrane of HEK-293 cells. Figure 8 depicts the primary feedback mechanism controlling PDE4 activity in response to prostaglandin-induced cAMP signals (at least in the first 5 min of stimulation). In this model PKA influences the time course of cAMP signals by stimulating PDE4 activity and by buffering cAMP (reducing free cAMP levels). AKAPs are required to concentrate PKA activity near PDEs. The equations used to describe this model are given below. Parameters and initial conditions are given in Table 1. (1) where, [N] is the total cAMP concentration, [cAMP] is the concentration of free cAMP, EAC is the cAMP synthesis rate, [PDE] is the concentration of unphosphorylated PDE4, kPDE is the cAMP hydrolysis rate constant for unphosphorylated PDE4, [PDEp] is the concentration of phosphorylated PDE4, kPDEp is the cAMP hydrolysis rate constant for phosphorylated PDE4, Km is the Michaelis constant for PDE4, [I] is the concentration of a competitive inhibitor such as rolipram, and KI is the inhibition constant. The phosphorylation of PDE4 is described by the equations below. (2) (3) where, kPKA is the rate constant of PKA-mediated phosphorylation, KI-PKI is the inhibition constant for the noncompetitive inhibitor PKI, kpp is the rate constant of phosphatase-mediated dephosphorylation, and [C] is the concentration of free catalytic PKA subunits. The time-dependent reduction in the rate of cAMP synthesis is approximated by (4) where, EAC0 is the initial rate of cAMP synthesis and τ is the time constant for the reduction in the synthesis rate. This reduction in cAMP synthesis may be due feedback regulation of AC activity or a small amount of receptor desensitization. The equations describing PKA activity are given below. (5) (6) (7) (8) (9) (10) (11) (12) (13) where, R represents the regulatory subunit of PKA with two cAMP binding sites, a and b. [RC] is the concentration of unliganded regulatory subunits bound to catalytic subunits, [RaC] is the concentration with cAMP bound to site a, [RbC] is the concentration with cAMP bound to site b, [RabC] is the concentration with cAMP bound to sites a and b. The rate constants for cAMP binding and unbinding to sites a and b are kfa, kfb, and kra, krb. The rate constants for the dissociation and association of the R and C subunits are kact and kdeact. Note that the dissociation and association of the R and C subunits are considered irreversible reactions. The concentration of free cAMP and activation of CNG channels are described as follows (14) (15) where I/Imax is the fractional current through CNG channels, K1/2 is the concentration of cAMP that elicits half-maximal current, and nH is the Hill coefficient.
Simulations of the model are shown in Fig. 9. All simulations were initiated by a step increase in AC activity, analogous to the rapid application of prostaglandins. Under control conditions the increase in AC activity initiates a transient increase in cAMP and CNG channel activity. The amplitude of the CNG channel response (Fig. 9A) and underlying free cAMP signal (Fig. 9B) are consistent with calibrated responses previously measured using the perforated patch technique (32). The decline in the cAMP signal is primarily due to PKA-mediated stimulation of PDE4 activity. The simulation predicts that the total cAMP concentration (free and bound) near the plasma membrane is ∼3-fold higher than the free cAMP concentration. This occurs because the free cAMP levels are significantly higher than the K1/2 of PKA (200 nM), even after 300 s. Note that the total cAMP concentration near the plasma membrane would be higher if the PKA concentration were increased (due to the higher buffering capacity discussed below). We next used the model to examine the effects of PDE and PKA inhibitors (Fig. 9, D–F and G–I, respectively). In the presence of 10 μM rolipram, a competitive PDE4 inhibitor, cAMP levels quickly increase, saturating CNG channels (Fig. 9D). The model predicts that total cAMP levels near the plasma membrane reach 20 μM in 5 min, ∼8-fold higher than control conditions. This is higher than the threefold increase in total cellular cAMP measured using enzyme immunoassays (Fig. 2). Factors that may contribute to these differences, and were not incorporated into the simulations, include mechanisms that segregate near-membrane and cytosolic cAMP signals (see Refs. 31, 32 for details), and incomplete equilibration of rolipram across the plasma membrane. Simulations using (intracellular) rolipram concentrations of 1–2 μM are consistent with the experimental observations presented in Fig. 2. In the presence of 20 nM PKI, a noncompetitive PKA inhibitor, cAMP levels reach a steady plateau in 5 min. This response demonstrates the importance of PKA-mediated stimulation of PDE4 activity in shaping the cAMP signal. Although basal levels of PDE4 activity are sufficient to prevent saturation of CNG channels (Fig. 9G) and to offset the rate of cAMP synthesis, they are not sufficient to cause the observed decline in cAMP levels.
We next sought to better understand the effects of PDE4 and PKA activity, as well as desensitization of cAMP synthesis rate (i.e., receptor desensitization or feedback inhibition of AC activity) on transient cAMP signals. We first examined the effects of altering kPDE from the control value of 0.11 s−1 (Fig. 10, A and B). A twofold reduction or increase in PDE4 activity dramatically altered the activation of CNG channels and the underlying free cAMP signals, either delaying the decline in free cAMP levels or dramatically reducing the amplitude of the response. These simulations clearly demonstrate the biological importance of maintaining precise control of PDE4 activity and the distribution of PDE4 within the cell. Altering the concentration of PKA also affected the time course of CNG channel activation and underlying cAMP signals (Fig. 10, C and D). Two- and fivefold increases in PKA concentration reduced the amplitude of the free cAMP signals, due to increased buffering and an increased phosphorylation rate, and created a significant lag in the activation of CNG channels. This simulation indicates that in cells with high local PKA concentrations, observed in a variety of cell types (13, 15, 26, 48), exposure to brief or subsaturating stimuli may activate PKA without activating lower affinity effectors such as endogenous CNG channels. Finally, a 10-fold reduction in τ had relatively small effects on the transient cAMP signals, primarily lowering peak and final cAMP levels (Fig. 10, E and F). This should not be surprising given that PKA-mediated stimulation of PDE4 activity has such profound effects on cAMP signals in this system. It is likely that processes such as receptor desensitization will have a greater impact on the time course of cAMP signals in systems in which stimulation of PDE activity does not dominate the response.
Previous studies have demonstrated that PKA phosphorylates PDE4D, and that phosphorylation triggers an increase in PDE4D activity (3, 41, 42). In separate studies, investigators have examined the functional role of PDE4 activity in shaping cAMP signals (1, 24, 32, 34, 49). Here we combine elements of each approach to more closely examine the contributions of hydrolysis and buffering in regulating free cAMP levels near the plasma membrane of HEK-293 cells. We accomplished this by examining the cellular mechanisms responsible for the decline in transient, prostaglandin-induced cAMP signals. We found that inhibition of PKA activity with H89 prevented the stimulation of PDE4 activity, as well as the decline of the transient cAMP signals in cell populations. To more definitively implicate PKA-mediated regulation of PDE4 activity, we examined the effects of application of H89 and PKI at known intracellular concentrations. These compounds prevented the decline in transient cAMP responses within expected concentration ranges for inhibition of PKA activity. We also examined the effects of an AKAP disruptor, Ht31, on PGE1-induced cAMP signals, and found that disruption of AKAPs significantly diminished the decline in transient cAMP responses. These results are consistent with recent studies that have shown HEK-293 cells express AKAPs 79 and 250 (20), and that AKAP 250 (gravin) plays a critical role in the subcellular localization of PKA and PDE4, as well as the kinetics of prostaglandin-induced cAMP signals (47). The effect of AKAP disruptors on these signals is likely to have occurred for two reasons: the reduction in catalytic subunit concentration lowered the rate at which PKA could phosphorylate PDE4, and the reduction in the regulatory subunit concentration lowered the local cAMP buffering capacity.
A study by DiPilato et al. (8) used fluorescent cAMP indicators based on EPAC to examine the dynamics of cAMP signals in HEK-293 cells. The EPAC-based sensors were targeted to different regions of the cell and cAMP responses to 10 μM isoproterenol, 10 μM PGE1, or 50 μM forskolin were observed. The agonist-induced cAMP responses reached steady plateaus within 4 min, whether the sensors were targeted to the plasma membrane or diffusely distributed to the cytosol. Interestingly, they observed a slow spatial spread of cAMP, consistent with our previous results (28, 31, 32). The differences in the kinetics of the observed responses may have occurred because the membrane localized EPAC sensors traffic to different regions of the plasma membrane than CNG channels, and thus measure a different subcellular pool of cAMP. Another possibility is that the HEK-293 cells used in the study by DiPilato et al. were inherently different. Consistent with the latter possibility, we routinely observe transient isoproterenol-induced cAMP signals using either CNG channel-based sensors or enzyme immunoassays (data not shown); however, the responses presented by DiPilato et al. reached a steady plateau, regardless of the subcellular localization of the EPAC-based sensor. Experiments need to be conducted in the same cells to reconcile potential differences between the responses monitored with different cAMP sensors.
Recently, Rochais et al. (37, 38) used CNG channels to examine cAMP PDE activity in rat cardiac myocytes. They found that exposure of cardiac myocytes to 100 nM isoproterenol triggered a transient cAMP response; however, the decline of the response was slower than those measured in HEK-293 cells. Extracellular application of H89 increased isoproterenol-induced and L-858051 (an analog of forskolin)-induced cAMP signals, as measured by CNG channels, but did not alter isoproterenol- or L-858051-induced stimulation of total cellular PDE3 or PDE4 activities. While the basis for the H89 effects is not fully understood, they were likely due, at least in part, to a PKA-mediated increase in PDE3 or 4 activity at the plasma membrane. They also found that cAMP signals triggered by different hormones were regulated by distinct subsets of PDE families present. The latter observation further implicates compartmentalization as the primary mechanism for the specificity of cAMP signals within cardiac myocytes.
The high resolution measurements of near-membrane cAMP signals presented here have allowed us to develop a plausible mathematical model of the cellular mechanisms underlying these responses. This model can be used to better understand cAMP signals measured in other systems. We have used the model to investigate the roles of PKA, PDE4, and processes such as receptor desensitization in shaping cAMP signals, by altering the concentration (PKA) or kinetics (PDE4 and receptor desensitization) of relevant enzymes. Two interesting predictions can be drawn from these simulations. First, processes such as receptor desensitization may not contribute significantly to the time course of transient cAMP signals if PKA-mediated stimulation of PDE4 activity is present. It remains to be determined if regulation of PDE activity and receptor desensitization act in concert to regulate cAMP signals, or if one process typically dominates the time course of the response. Second, the model predicts that nonuniform concentrations of PKA may buffer cAMP to different extents in different subcellular compartments. Beautiful examples of nonuniform PKA distribution caused by AKAPs are presented in Refs. 26 and 50. AKAPs also help to generate nonuniform concentrations of other proteins, including PDE4 (10, 48). High levels of localized cAMP buffers (e.g., 5 μM PKA) and PDE4 would allow cAMP signals to activate PKA without activating EPAC or native CNG channels. In this scenario, EPAC and CNG channels would be activated after prolonged stimulation of AC (after the buffers had been saturated), or when PDE activity is inhibited. This hypothesis is particularly intriguing given the recent discovery of a signaling complex that contains mAKAP, PKA, PDE4D3, and EPAC1 (9). The measurement of local PKA concentration, as well as the concentration of other proteins within signaling modules, is an important area of future research. Such data are necessary if we are to determine whether localized PKA concentrations do indeed contribute to the segregation of cAMP-mediated cellular responses, and to understand how information is transmitted through the cAMP signaling pathway.
This work was supported by National Institutes of Health Grants R01-HD-20788 (to M. Conti), R01-EY-09275 (to J. W. Karpen), and R01-HL-074278 (to T. C. Rich), and the State of Texas Advanced Technology Program Award 011618-0106-2003 (to T. C. Rich).
We thank Drs. Richard Clark, Carmen Dessauer, and Jeffrey Frost for valuable discussions about this research, and Dr. Troy Stevens for helpful comments on the manuscript.5
The costs of publication of this article were defrayed in part by the payment of page charges. The article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
- Copyright © 2007 the American Physiological Society