Cyclic AMP Levels, Adenylyl Cyclase Activity, and Their Stimulation by Serotonin Quantif ied in Intact Neurons

Ligand-gated ion channels typically exhibit binding affinities for their ligands within the concentration range they normally encounter. For the cAMP-gated Na⁺ currents (I_Na,cAMP) of molluscan neurons, maximal activation is not achieved at cAMP concentrations below 100 μM (Sudlow et al., 1993). Such high, naturally occurring concentrations of cyclic nucleotides have been inferred from conventional biochemical measurements in other cell types possessing cyclic nucleotide–gated cation currents: photoreceptor cells of retina and peripheral sensory neurons of olfactory epithelium in vertebrates (Pace et al., 1985; Stryer, 1986; Breer et al., 1990; Koutalos et al., 1995). These concentrations are many times in excess of those causing maximal activation of cAMP-dependent protein kinase (Beavo et al., 1974). In molluscan neurons, cAMP mediates both short- and long-term stimulation of excitability in both sensory neurons (Siegelbaum et al., 1982; Ocorr et al., 1985; Walsh and Byrne, 1989; Goldsmith and Abrams, 1991; Byrne et al., 1993) and neurons of motor networks (Gillette et al., 1982; Gillette and Gillette, 1983; Connor and Hockberger, 1984; Gillette and Green, 1987; Kirk et al., 1988; Kehoe, 1990; Huang and Gillette, 1991; Funte and Haydon, 1993; Price and Goldberg, 1993; Sudlow and Gillette, 1995); thus, it is of appreciable interest to know the levels of cAMP prevailing in both resting and stimulated central neurons, and the levels of synthetic activity that maintain them.

Using I_Na,cAMP as a reporter of intracellular cAMP concentrations, we measured cAMP concentrations and adenylyl cyclase (AC)¹ activities in neurons in the locomotor network of the predatory gastropod Pleurobranchaea californica, where we compared the values for resting state with those obtained where AC was stimulated by the neuromodulator serotonin, 5-hydroxytryptamine (5-HT). Our results indicate that stimulated levels of cAMP are indeed in the range of tens of micromolar, and that these are achieved through correspondingly high levels of AC stimulation.

Neurons of the “G” group of the pedal ganglia of the sea slug P. californica were used in all experiments. These cells are putative locomotor neurons in which I_Na,cAMP is stimulated by 5-HT through activation of AC (Sudlow and Gillette, 1995). Specimens (80–500 g) of Pleurobranchaea were obtained from Sea Life Supply (Sand City, CA) and Pacific BioMarine (Venice, CA) and maintained in 14°C artificial sea water. Animals were anesthetized by chilling, the pedal ganglia were removed, and axotomized somata (110–240 μm diameter) were isolated and prepared for voltage clamping as previously described (Sudlow and Gillette, 1995). Artificial Pleurobranchaea saline consisting of (mM) 420 NaCl, 10 KCl, 25 MgCl₂, 25 MgSO₄, 10 CaCl₂, 10 MOPS buffered with NaOH, final pH 7.5, 14°C was constantly perfused over the cells. Cells were clamped via two-electrode voltage clamp amplifier with total cellular current reported by virtual ground amplifier. The voltage-sensing electrode was filled with 3 M KCl. A double-barreled electrode was used for current passing and cAMP iontophoresis. The current passing barrel was filled with 3 M KCl and the second barrel was filled with 200 mM cAMP (adenosine 3′:5′ cyclic monophosphate free acid; Sigma Chemical Co., St. Louis, MO), 200 mM KOH, and 20 mM Tris, pH 7.4. Resistances of both double-barreled and single pipettes were lowered by breaking back the tips against a clean capillary tube. Typical resistances for voltage and current electrodes were 4–12 and 35–80 MΩ, respectively, for the cAMP iontophoresis barrel. The double-barreled electrode was positioned at the center of the soma with a calibrated micromanipulator, as estimated from soma diameter measured with a calibrated eyepiece micrometer. All current recordings were done at a holding potential of −50 mV. Iontophoretic injections of cAMP were made with a programmable constant current source (260; WPI, New Haven, CT). Variation of iontophoretic currents over multiple injections at the same programed delivery level was negligible.

A transport number for cAMP iontophoresis was measured by radioimmunoassay. cAMP was iontophoresed from double-barreled electrodes (current barrel resistance 2–8 MΩ) into the mouth of a collecting siphon tube (PE 10 polyethylene tubing) flowing at 20 μl·min⁻¹. Samples were collected in polypropylene microcentrifuge tubes and frozen for subsequent analysis. Saline in the reservoir from which the siphon collected was designed to mimic cytoplasmic cation composition of molluscan neurons, and contained (mM) 400 KCl, 50 NaCl, and 1 Tris, pH 7.5. Frozen samples were dried in a vacuum centrifuge (Speed-Vac; Savant Instruments, Inc., Farmingdale, NY) and radioimmunoassayed as below. cAMP delivery from three electrodes resulting from −20 and −100 nA, 50-s iontophoretic currents yielded a linear relation extrapolating to zero; an average transport number of 0.11 (0.01 SEM) was calculated from the −100 nA samples (5.79 pmol cAMP; 0.61 SEM).

The G lobes of pedal ganglia were excised with small scissors as near-spherical clumps of cell bodies whose volumes were estimated from the approximate diameters measured by a calibrated ocular micrometer. The isolated G lobes were allowed to recover for 1 h at 14°C, and then perfused either with normal saline or saline containing 100 μM 5-HT (serotonin creatinine sulfate complex; Sigma Chemical Co.) for 5 min, after which a cold (−20°C) antifreeze solution of 1:1 (vol/vol) propylene glycol:2 M NaCl was flushed into the dish (Greenberg et al., 1987). The frozen tissue was placed in a −20°C freezer for 1–2 h, and then homogenized in a glass microhomogenizer (Radnoti Glass Technology, Inc., Monrovia, CA) in 50 μl ice-cold 65% ethanol with 10 turns of the pestle. The homogenate was transferred to a clean microcentrifuge tube and the homogenizer was washed twice with 50-μl rinses of ice-cold 65% ethanol into the microcentrifuge tube. Homogenate and rinses were mixed and immediately aliquotted for both cAMP and protein assays. Aliquots were frozen at −20°C and dried on a vacuum centrifuge. Freeze-dried aliquots were stored at −20°C until assayed. Protein concentrations were determined via the Bradford method (Bradford, 1976).

cAMP was measured by scintillation proximity assay (SPA), using a modified RIA protocol (Amersham Corp., Arlington Heights, IL). The SPA measures cAMP via competition between cAMP and adenosine 3′:5′-cyclic phosphoric acid 2′-O-succinyl-3-[¹²⁵I]iodotyrosine methyl ester for an antibody/scintillant microbead system. The dried samples were dissolved in 0.5 ml of 0.05 M acetate buffer immediately before assay. Samples were processed under the nonacetylation protocols for the kit. The range of sensitivity for the nonacetylation protocols of the kit was 0.2–12.8 pmol/ assay tube. Aliquots (100 μl) of the samples were transferred to polypropylene scintillation vials (Sarstedt, Inc., Newton, NC). SPA analysis of the samples was performed as per package instructions and assay vials were counted on a β-scintillation counter. Data were adjusted for percent recovery, typically 90%.

5-HT (serotonin creatinine sulfate complex; Sigma Chemical Co.) was dissolved in Pleurobranchaea saline and pH was adjusted to 7.5 with NaOH. A gravity-fed perfusion pipette was placed upstream of the voltage-clamped somata.

Each soma was internally calibrated for its I_Na,cAMP responses to iontophoretically injected cAMP. In addition to the I_Na,cAMP amplitudes, two additional sets of measures were made during steady state injections of cAMP. First, for the pedal ganglia cells, depolarizing voltage command steps to inactivate I_Na,cAMP were delivered from −50 (holding) to +10 mV for 100 ms. The degree of inactivation was measured as the tail current (I_tail) representing the net decrease in I_Na,cAMP amplitude measured 1 s after the voltage step delivered during steady state cAMP injections or 5-HT application (Gillette and Green, 1987; Huang and Gillette, 1993; Sudlow and Gillette, 1995). The second measure examined the exponential decay rate of I_Na,cAMP after 5-s pulsed iontophoretic injections of cAMP superimposed on the steady state injections. The data obtained during the decay were digitized and were fit by a least-squares algorithm to a time-dependent exponential decay function (Y = X exp (−t   · k_h)) to obtain the exponential term k_h where k_h is in the units of 1/s and t is time in s. The decay rate decreased as steady state iontophoretic current delivery increased. The basal levels of I_Na,cAMP could be calculated from either the I_tail or k_h values obtained during the absence of exogenous cAMP injections using the measured slope of the inactivation/ I_Na,cAMP relation, or from the slope of the k_h/I_Na,cAMP relation.

Exponential decay slopes (k_h) of I_Na,cAMP after 5-s pulsed injections of cAMP are directly proportional to the rate of change in [cAMP]. The decrease in [cAMP] is directly attributable to degradation by phosphodiesterase (PDE) (Huang and Gillette, 1991). While some cellular binding of cAMP may occur, buffering effects are not evident and are presumed negligible. The dose–response relationship between k_h and steady state I_Na,cAMP responses was used to calculate a k_h for an I_Na,cAMP response to injected cAMP of equal amplitude to the 5-HT-induced I_Na,cAMP (I_5-HT). This equivalent k_h and the iontophoretic current equivalent for I_5-HT were then used in Eq. 5 to calculate a cAMP concentration at the submembrane space ([cAMP]_memb). The k_h and [cAMP]_memb obtained during 5-HT stimulation were used in Eq. 6 to calculate AC activity.

Iontophoretic equivalents for basal and 5-HT-stimulated I_Na,cAMP responses were calibrated on a molar basis by Eq. 1, which describes the rate of release of cAMP from an iontophoretic pipette tip (Purves, 1981):

where q_i is the iontophoretic molar flux (mol · s⁻¹), I is the iontophoretic current, n is the transport number for cAMP empirically derived for these experimental conditions, z is the valence, and F is the Faraday constant. The total amount of cAMP delivered during a steady state iontophoretic injection is equal to q_i·t, where t is time in s.

Two additional processes are involved in the movement of cAMP out of the iontophoretic pipette: diffusional leak and bulk flow, which can both be significantly affected by the diameter of the pipette tip. Passive diffusion can be described by the relation (Purves, 1981):

where q_d is the diffusional leak (mol·s⁻¹), D is the apparent diffusion coefficient (3.3 ·10⁻⁶ cm² · s⁻¹) for cAMP in Pleurobranchaea neurons (Huang and Gillette, 1991), C_pip is the concentration of cAMP in the pipette (0.2 M), and θ is the included angle in radians of the pipette tip from the center axis of the pipette; typically for our pipettes the angle was 1.5°. The term r_pip is the inside radius (cm) of the pipette. We assumed that breakage of the double-barreled electrodes broke both barrels equally. Effective tip diameters were calculated from resistances for the current passing barrels of the double-barreled electrodes using the relation (Purves, 1981):

where σ describes the conductivity of the 3 M KCl electrolyte of the voltage clamp current barrel, nominally 0.26 S·cm⁻¹ (Purves, 1981); and R_ue is the pipette resistance in ohms. Using this relationship, the range of the resistances of the current passing electrodes corresponded to tip diameters of 0.08–0.23 μm.

Bulk flow arises from hydrostatic pressure effects imposed at the pipette tip by gravity by the relation (Purves, 1981):

where q_h is the hydrostatic bulk flow (mol·s⁻¹), p is the viscosity (1,000 kg·m⁻³), g is gravitational attraction (9.8 m·s⁻²), h is the height of cAMP solution in the iontophoretic barrel (m), and η is the viscosity of the cAMP solution (0.001 Pa·s). The terms θ, C_pip, and r_pip have the same meanings as above. Diffusional leak and hydrostatic bulk flow of cAMP from the cAMP iontophoresis barrels were calculated to have contributed to <3 fmol·s⁻¹ cAMP under these conditions given the pipette diameters involved (Krnjević et al., 1963; Purves, 1981). In radioimmunoassays of sample aliquots taken after 300 s of diffusion with low resistance iontophoretic electrodes, cAMP levels did not register above the kit's detection limits (0.2 pmol/aliquot).

The steady state distribution of a solute (such as cAMP) released from a point source (the injection electrode) in a finite spherical space with an impermeable boundary (the axotomized neuron soma), in which a first-order decay occurs, is described by Eq. 5 (Purves, 1976, 1977):

where C(r) is cAMP concentration at radius r, q_i is the iontophoretic injection flux described by Eq. 1, r_n is the radius of the cell membrane, k_h is the relaxation rate constant for I_Na,cAMP elicited by I in Eq. 1, and D has the same value as above. We assumed the simplest cases where PDE activity is distributed homogeneously in the cytoplasm and there are no significant internal diffusion barriers (see Huang and Gillette, 1991). The exponentially decreasing gradient from pipette tip to cell membrane described by Eq. 5 is largely influenced by the diameter of the neuron and PDE, which is reflected in k_h. Eq. 5 was simplified from its original form since the radius of the release point has no significant impact on the computations as long as the inside diameter of the pipette is <4 μm. In practice, our pipette tip diameters were <1 μm.

We related quantitative cAMP injections to steady state I_Na,cAMP, its inactivation characteristics, exponential decay rates, and measurements of current saturation. I_Na,cAMP reaches a steady state condition during tonic iontophoretic cAMP injections (Fig. 1). At nonsaturating levels, the amplitude of elicited I_Na,cAMP increases linearly with the iontophoretic current (Figs. 1 and 2,A). Although most experiments were performed in the nonsaturating range of the iontophoretic current/I_Na,cAMP dose–response function, some experiments were designed to specifically examine the saturability of I_Na,cAMP by iontophoretic current. The I_Na,cAMP responses did saturate at high current (>−200 nA) iontophoretic injections of cAMP (Fig. 3). These iontophoretic current/ I_Na,cAMP saturation measures were not typically performed in most experiments, due to adverse long term consequences on the viability of the cell and on diminishing amplitude and lengthening duration of I_Na,cAMP after prolonged saturating injections of cAMP.

I_Na,cAMP in pedal neurons is inactivated by Ca²⁺ influx caused by depolarizing voltage steps (Fig. 1); after such a step, inactivation is manifested as a slowly decaying tail current (I_tail; Gillette and Green, 1987; Huang and Gillette, 1993). When brief depolarizing voltage steps are delivered to otherwise unstimulated cells, small tail currents are observed resulting from inactivation of basal I_Na,cAMP. The I_tail amplitude increases linearly with the inward current induced during either tonic iontophoretic injections of cAMP or bath application of 5-HT, indicating the origin in I_Na,cAMP (Figs. 1 and 2 B).

Exponential decay slopes (k_h) of I_Na,cAMP after 5-s pulse injections of cAMP were also dependent on the steady state level of I_Na,cAMP in the cell (Fig. 1), with k_h decreasing with increasing injections of cAMP (Fig. 2 C). This relationship between k_h and steady state I_Na,cAMP was linear when the iontophoretic injections were below saturation levels for I_Na,cAMP. The relaxation rate k_h is highly sensitive to PDE inhibitors (Aldenhoff et al., 1983; Connor and Hockberger, 1984; Huang and Gillette, 1991) and is a quantitative index of PDE-mediated degradation of cAMP (Huang and Gillette, 1991). PDE inhibitors, such as 3-isobutylmethylxanthine, cause dose– dependent decreases in the decay rate k_h and increases in the cAMP levels in the soma without change in adenylyl cyclase activities (Sudlow and Gillette, manuscript in preparation).

The steady inward current stimulated by 5-HT (I_5-HT) occludes the I_Na,cAMP response to iontophoretic pulsed injections of cAMP (Figs. 1 and 2 D). I_5-HT in the pedal neurons is largely composed of I_Na,cAMP, whose levels can be calculated from inactivation (Sudlow and Gillette, 1995) or measures of k_h saturation (see further). The degree of occlusion was also indicative of the degree of saturation of I_Na,cAMP by iontophoretic current. As the iontophoretically elicited I_Na,cAMP approached saturation, the degree of occlusion also increased.

Since I_Na,cAMP is a direct function of the cAMP concentration, the amplitude of the current is the balanced result of the activities of the synthetic and degradative enzymes, AC and PDE, respectively (Huang and Gillette, 1991; also see Hodgkin and Nunn, 1988). Knowledge of the cAMP concentration and of the activity of at least one of the enzymes would allow the calculation of the remaining enzyme's activity at steady state. However, in the living cell, unlike the biochemist's reaction vessel, such a calculation must take into account the compartmentalization and geometry of the cell and the kinetics of cAMP diffusion and degradation. In the neuron, AC is bound at the plasma membrane, while PDE is distributed in the cytoplasm in both soluble and membrane-bound fractions (Fell, 1980; Shakur et al., 1993; Bolger, 1994). Thus, we took into account the existence of a gradient of cAMP concentrations. During neurotransmitter stimulation of the soma, this cAMP concentration gradient is highest next to the membrane and falls exponentially with distance towards the center of the cell (Fell, 1980; Bacskai et al., 1993; Shakur et al., 1993; Bolger, 1994).

AC activity was calculated from the I_Na,cAMP records obtained before and during bath applications of 5-HT. The responses of the cell were calibrated using three relationships based upon direct experimental measures: (a) the dose–response relation of tonic cAMP iontophoretic current to the induced steady state I_Na,cAMP (see Fig. 2,A); (b) the dose–response relation of steady state I_Na,cAMP current amplitudes to depolarization- induced inactivation (see Fig. 2,B), which allowed estimation of I_Na,cAMP in the absence of cAMP injection; and (c) the dose–response relation of the steady state I_Na,cAMP to k_h (see Fig. 2,C) obtained during steady state I_Na,cAMP, which also allowed for the estimation of basal I_Na,cAMP (Huang and Gillette, 1991). Resting and 5-HT-stimulated levels of I_Na,cAMP were calculated using the resting and 5-HT-stimulated inactivation measures and the linear regression coefficients of the inactivation/steady state I_Na,cAMP relation (see Fig. 2,B). The k_h values associated with the steady state I_Na,cAMP equivalents for resting and 5-HT-stimulated states were determined from the I_Na,cAMP equivalents and the linear regression coefficients of the k_h/steady state I_Na,cAMP dose–response relation (see Fig. 2,C). The steady state I_Na,cAMP equivalents were translated into iontophoretic cAMP injection equivalents using the linear regression coefficients of the dose–response relation of tonic injection current to induced I_Na,cAMP (see Fig. 2 A). These iontophoretic equivalents, along with their equivalent k_h values, were adjusted for diffusion and hydrolysis by Eq. 5 to estimate [cAMP]_memb. Last, AC activities were calculated from the [cAMP]_memb and the k_h during 5-HT-stimulated state using Eq. 6. We derived Eq. 6 (appendix) to explicitly describe the active synthesis, degradation, and inward diffusion of an enzymatic product from the boundary of an impermeable sphere:

where C(r) is [cAMP] at radius r, taken here as the compartment 100 nm subjacent to the cell membrane. Q_AC is the flux of the membrane-bound AC expressed in terms of mol·s⁻¹. The variables r, r_n, k_h, and D have the same meanings as in Eq. 5. The equation was derived with the assumption that PDE is homogeneously distributed in the soma. After calculating [cAMP]_memb with Eq. 5, Eq. 6 is solved for Q_AC. With the soma diameter, AC activity is normalized to the surface area of the soma (mol·s⁻¹·cm⁻²) to allow ready comparison of different cells.

The results of measurements on 30 neurons are summarized in Table I. In unstimulated cells, the distribution of basal [cAMP]_memb was markedly skewed to lower values with a few rare higher values (Fig. 4); AC activities were distributed similarly. Stimulation of AC by 5-HT shifted the distribution of responses towards markedly higher values (Fig. 4). The non-Gaussian distributions of [cAMP]_memb and AC activities contributed to wide variation among the calculated means and standard deviations. When cAMP concentrations and AC activities in bilaterally homologous neurons from the pedal ganglia of the same animal were examined (see Table I, cells A and B, E and F, and K and L), the levels were very comparable. However, when cAMP concentrations and AC activities in different, nonhomologous neurons of the same animal were compared (see Table I, cells F vs. G, I vs. J, and P vs. Q), levels varied substantially. Thus, variability of the cAMP and AC measures is likely to arise from phenotypic variation in the cell population under study. 5-HT stimulation augmented [cAMP]_memb from 1.16- to 82.60-fold, averaging 23.35- ± 21.96-fold (μM ± SD), and AC from 1.17- to 89.68-fold, with an average stimulation of 18.56- ± 18.79-fold. The average resting [cAMP]_memb was 3.64 ± 8.59 (μM ± SD). Adjusting for the PDE-induced gradient in cell bodies yields an integrated average whole-cell concentration of 2.54 ± 2.61 μM. 5-HT stimulation raised [cAMP]_memb to 27.68 ± 30.12 μM and whole-cell integrated [cAMP] to 20.60 ± 23.61 μM. Similarly, resting AC activities averaged 2.44 ± 5.26 × 10⁻¹² mol·s⁻¹·cm⁻² and were elevated by 5-HT stimulation to 15.96 ± 16.21 × 10⁻¹² mol·s⁻¹·cm⁻². When these AC activities are expressed in terms of cell volume, as μM·s⁻¹ (micromol·liter⁻¹·s⁻¹), basal AC was 0.83 ± 1.64 μM·s⁻¹ and stimulated AC rates (basal + 5-HT stimulated) were 5.55 ± 6.92 μM·s⁻¹.

Radioimmunoassays of basal levels of cAMP in the G lobes yielded an average value of 0.82 ± 0.45 × 10⁻¹² mol·(μg protein)⁻¹ (μM ± SD; n = 2). We used a presumedly maximal stimulatory concentration of 5-HT (100 μM; see Fig. 7) to observe the upper end of cAMP production. Such stimulation elevated cAMP to 9.08 ± 1.95 × 10⁻¹² mol·(μg protein)⁻¹ (n = 5) within a range of 6.38–11.46 × 10⁻¹² mol·(μg protein)⁻¹. The protein concentration per unit volume, measured by Bradford assay, was 8.5 μg·μl⁻¹. With this conversion factor, the electrophysiological measures of whole cell [cAMP] (Table I) could be directly compared with the radioimmunoassays. The radioimmunoassayed levels of unstimulated G lobes were 6.97 ± 3.84 μM and levels during stimulation with 100 μM 5-HT were 77.15 ± 16.86 μM within a range of 54.2–97.38 μM. This compares with the electrophysiological measures of [cAMP] in unstimulated somata of 3.64 ± 8.59 μM (n = 30), within a range of 0.08–28.88 μM, and 10 μM 5-HT-stimulated whole cell [cAMP] of 20.60 ± 23.61 μM, within a range of 1.73–96.90 μM. Results of both biochemical and electrophysiological assays agree that 5-HT stimulates high levels of cAMP.

We next examined the apparent binding affinity of cAMP in activating I_Na,cAMP. The apparent binding affinity for cAMP is an inverse function of intracellular [Ca²⁺] in the G cells (Huang and Gillette, 1993), and is a minimum estimate of actual binding affinity. The I_Na,cAMP elicited for each tonic iontophoretic injection and the [cAMP]_memb determined for that injection were fit by least-squares analysis to the modified Hill equation (Huang and Gillette, 1991):

where I_max is the maximum available I_Na,cAMP and K_c is the apparent cAMP dissociation constant for the I_Na,cAMP channels. The average I_max was 20.92 ± 16.53 nA (n = 17) within a range of 6.84 –90.62 nA. The average K_c for the I_Na,cAMP current was 18.05 ± 8.95 μM (n = 17) within a range of 4.32–34.05 μM. For experiments designed to examine the interaction between 5-HT and I_Na,cAMP, the calculated [cAMP]_memb and nonsaturating I_Na,cAMP levels were fit by a least-squares algorithm of Eq. 7. Additional experiments were designed to specifically examine the entire dose–response relation between iontophoretic current and I_Na,cAMP (see Fig. 3). The relation, presented in terms of the [cAMP]_memb/I_Na,cAMP response, was fitted well by Eq. 7 (Fig. 5). The I_max and K_c for the data in Fig. 3, calculated as per above, were 16.86 nA and 22.12 μM, respectively. These values are consistent with cAMP dissociation constants for I_Na,cAMP channels thought to be at least several tens of micromolar or greater (Aldenhoff et al., 1983; Green and Gillette, 1983; Connor and Hockberger, 1984; Sudlow et al., 1993).

The net increase in AC activation stimulated by 5-HT was inversely proportional to resting AC rates (Fig. 6). In general, cells that exhibited comparatively high basal AC rates (see Table I) showed appreciably less stimulation by 5-HT than cells with low basal activity. Collectively, these data indicate that cyclase activity varies in its basal levels of activation among different cells and preparations, perhaps due to effects of dissection, endogenous neuromodulators, or other unknown factors.

5-HT elicited similar concentration-dependent effects in pairs of bilaterally homologous neurons (Fig. 7). Measured values of AC activation with increasing 5-HT concentrations for bilateral homologs (Table I, cells K and L) were similar, reaching a plateau phase above 10 μM 5-HT (Fig. 7).

The present findings document high average resting levels of submembrane cAMP in unstimulated neurons, in excess of 3.64 μM, within a wide range of 0.13–44 μM. The whole cell levels of cAMP were 2.54 μM within a range of 0.07–50 μM, which corresponded well with the radioimmunoassay measures of cAMP in Pleurobranchaea G lobes homogenates of 6.97 μM. These values are also similar to the results of radioimmunoassays of homogenates of large peripheral effector neurons of Aplysia, neurons similar in function to those studied here, that yielded resting levels of cAMP of 1–6 μM (Hockberger and Yamane, 1987). These values in Pleurobranchaea, even at the lower limit, are considerably higher than the resting levels of <50 nM in the somata of cultured Aplysia sensory neurons, as estimated from cAMP binding-induced fluorescence changes in fluorophore-labeled subunits of PKA (Bacskai et al., 1993). This is unlikely to result from a species difference between Aplysia and Pleurobranchaea, since our unpublished observations on identified effector neurons of Aplysia (R2 and L10 of the abdominal ganglion) exhibiting I_Na,cAMP show that they require a similar range of iontophoretic current to activate the current. The differences in values for [cAMP] between sensory and effector neurons may well be adaptively related to the expression of I_Na,cAMP in the effector neurons. I_Na,cAMP is probably not significant in the sensory neurons; they are well studied and a prominent I_Na,cAMP would doubtless have been reported previously. Thus, these results suggest that cAMP levels tend to be higher in neurons that express I_Na,cAMP than in those that lack it. This interpretation is consistent with observations on vertebrate retina and olfactory tissues, where cyclic nucleotide levels in cells expressing similar cGMP- or cAMP-gated cation currents attain similarly high values.

In 5-HT-stimulated somata the average whole cell cAMP levels rose to nearly 21 μM. Because of the concentration gradient for cAMP caused by asymmetric distribution of AC and PDE, the actual average submembrane values exceeded 27 μM and in several cases approached 100 μM (Table I). Observations on single I_Na,cAMP channels in excised patches showed that the probable cAMP dissociation constant was probably in excess of some tens of micromolar (Sudlow et al., 1993), a conclusion in accord with observations on the macro-current (Aldenhoff et al., 1983; Green and Gillette, 1983; Connor and Hockberger, 1984; Sudlow et al., 1993). In the present measures, we found that the average dissociation constant for cAMP binding in activation of I_Na,cAMP was about 18 μM, within a range of 4.32–34.05 μM. The wide variation in the apparent binding affinity for cAMP is likely to arise in large part from the fact that it is a function of intracellular [Ca²⁺] in the G cells, which acts like a competitive inhibitor of cAMP binding to the channel (Huang and Gillette, 1993); other modes of channel regulation, such as phosphorylation (see Wilson and Kaczmarek, 1993), are also possible. The present observations support the expectation for the relatively low binding affinity for cAMP in activation of the macro-I_Na,cAMP in the intact neuron. They also show that the high endogenous levels of cAMP needed to activate the current are actually attained during stimulation with the natural neuromodulator.

The system of equations presented in this study incorporates an assumption of homogeneous distribution of PDE in the soma. This assumption stems from observations of Huang and Gillette (1991), who varied the depth of the iontophoretic pipette and examined the exponential decay rates of I_Na,cAMP after pulse injections of cAMP; they found no variation of k_h with depth. If PDE were significantly excluded from the nuclear core of the soma, and PDE were concentrated in the rind of the cytoplasm much like the distribution of regulatory subunit of PKA observed by Bacskai et al. (1993), then the exponential decay after a pulse injection of cAMP should have slowed as the pipette penetrated further to the center of the cell, where the PDE-free zone would have acted as a cAMP reservoir. Instead, the modeling of the time course of the response of I_Na,cAMP to a pulse injection was best fit by a homogenous distribution of PDE, where depth and PDE activity accounted for the response's characteristics and time course (Huang and Gillette, 1991). We are not aware of direct immunocytochemical documentation for the distribution of molluscan PDE in the soma. If in the future it should be found that Pleurobranchaea G neurons have substantial regions of the soma devoid of PDE, the results presented in this report will be underestimations of the cAMP levels of the cells in Table I. Conversely, a second nonhomogenous distribution of PDE could involve a thin region of high PDE activity subjacent to the neurolemma. Careful examination of the performance of the equations indicated that as long as the region of high PDE activity (i.e., 2·k_h) involved <14% of the total volume of the cell, the consequences to the cAMP calculations amounted to <1% variations in the results. The equations function not only due to the effects of k_h and diffusion, but through the distance that the cAMP has to diffuse. If the region of high PDE activity is only a few microns thick near the neurolemma, the consequences to the computations are negligible.

The [cAMP] and AC activities presented in Table I exhibit fairly large variation reflected in the means and standard deviations calculated for the sample. Some of the variability during resting measures can be accounted for by the extremely high levels of cAMP and AC in the two cells T and V, which appeared to be maximally activated in the basal state. The high levels associated with those cells are indicative of significant AC activation at the time of the experimental measures. During 5-HT stimulation, the measures associated with cells R and DD contributed to the variance of cAMP and AC measures. The levels of cAMP associated with cells R and DD, while high, are similar to the radioimmunoassay measures of cAMP obtained during 5-HT treatment, and thus appear to be within physiological range. It is highly unlikely that errors in the measures of k_h would account for the large sums of squares associated with the measures in Table I. We examined the performance of the equations and found that for a given model cell, iontophoretic current, and k_h, k_h would have to be underestimated by >60% to reach cAMP values in excess of the mean [cAMP]_memb, 27.6 μM. Typically, <2% variability was observed in repeated exponential slope measurements of k_h of the same I_Na,cAMP response to cAMP pulse injection. The production of cAMP during 5-HT treatment falls into two major categories, cells that respond with cAMP levels between 2.97 and 13.91 μM with a mean and standard deviation of 7.63 ± 3.82 μM (n = 18) and those cells that respond with cAMP production between 31.46 and 119.35 μM with a mean and standard deviation of 57.97 ± 26.92 μM (n = 12) (P < 0.0001, Mann-Whitney U test). It appears that different Pleurobranchaea G neurons vary greatly in cAMP production in response to 5-HT and it is this phenotypic variation that accounts for the variability of the individual cell measures. This was borne out by the observations on nonhomologous neurons from pedal ganglia of the same animal (Table I: cells F vs. G, I vs. J, and P vs. Q).

The radioimmunoassay measures of cAMP of whole Pleurobranchaea G lobe homogenates were higher than most of the cellular measures of Table I. These homogenates involved not only somata, but neuropil as well. cAMP levels in dendrites have been shown to be significantly higher than that observed in the somata in Aplysia sensory neurons (Bacskai et al., 1993). The higher levels of cAMP measured in the RIA as compared with the cellular measures may reflect, in part, the contributions of a higher neuropilar stimulation of cAMP production by 5-HT. Additionally, the neurons of the G lobes of the animals used in the RIA as a group may have been prone to produce cAMP at higher levels, like that observed in Table I, cells R, Y, and DD.

The average 18-fold stimulation of AC activity, and the increases shown by some neurons up to nearly 90-fold, are appreciably higher than values reported for in vitro measures of molluscan nervous tissue, which may vary from 0.75- to 7-fold in homogenates of Aplysia sensory neurons (Yovell et al., 1987; Eliot et al., 1989; Abrams et al., 1991; Goldsmith and Abrams, 1991; Yovell and Abrams, 1992). These differences could result from variation in AC activity among cells on the basis of whether or not they express I_Na,cAMP. Alternatively, or in addition, the considerable differences between in vitro and in vivo stimulation may well arise from the disruption of the intracellular environment inherent to in vitro assays. In particular, homogenization may greatly reduce the capacity for AC activity by dilution of the G protein cofactors and disruption of the tubulin that interacts with the G proteins. This interpretation is suggested from observations on adrenergic stimulation of AC in an assay system where cells are detergent permeabilized, but remain otherwise intact; AC stimulation in these cases can exceed 15-fold (Rasenick et al., 1993; Popova et al., 1994).

Nucleotide cyclase activity in vertebrate photoreceptors, which use cyclic nucleotide–gated currents in sensory transduction, appears to be regulated 5–40-fold between light- and dark-adapted conditions, similar to the range of stimulation of AC observed here (Pace et al., 1985; Koutalos et al., 1995). Use of the protein per unit volume conversion factor allows comparison of the activities of AC in Pleurobranchaea and cyclase activity from vertebrate olfactory receptors and photoreceptors. Unstimulated levels of Pleurobranchaea AC were 5.86 ± 2.12 × 10⁻⁹ mol·min⁻¹·(mg protein)⁻¹. 5-HT-stimulated levels of AC were 39.18 ± 7.62 × 10⁻⁹ mol·min⁻¹·(mg protein)⁻¹. These ranges of cellular cAMP concentrations and cyclase activity are similar to vertebrate tissues that also express cyclic nucleotide–gated currents (Pace et al., 1985; Stryer, 1986; Breer et al., 1990; Koutalos et al., 1995).

Cyclic nucleotide–gated currents have been identified in a wide variety of tissues including vertebrate photoreceptors, vertebrate and invertebrate olfactory receptors, cochlear hair cells, cardiac myocytes, sperm cells, and cultured hippocampal neurons (Haynes and Yau, 1985; Fesenko et al., 1985; Nakamura and Gold, 1987; Firestein et al., 1991; Zufall et al., 1991; Kolesnikov et al., 1991; DiFrancesco and Tortora, 1991; Hatt and Ache, 1994; Weyand et al., 1994; Leinders-Zufall et al., 1995). Many other large, identified neurons of gastropod central nervous systems also express I_Na,cAMP (Liberman et al., 1975; Green and Gillette, 1983; Aldenhoff et al., 1983; Kononenko et al., 1983; Connor and Hockberger, 1984, 1985; Gillette and Green, 1987; McCrohan and Gillette, 1988; Kehoe, 1990; Huang and Gillette, 1991, 1993; Price and Goldberg, 1993; Sudlow et al., 1993). For other cell types where cyclic nucleotide–gated ion current is prominent, the analytical approach introduced here with suitable modifications may complement conventional measures of cyclic nucleotides, particularly for the case of single cells that are too small to assay by presently available biochemical techniques. The general method may be adaptable by the use of caged cyclic nucleotides and several other possible indices of baseline current. Introduction of cyclic nucleotide–gated channels either genetically or directly to act as reporter ion channels may be practical in many types of cells. Aside from studies of neuromodulator-neurohormone actions as performed here, these methods are applicable to inquiries of receptor-G protein-AC coupling, interactions of other intracellular messenger pathways with cAMP metabolism, and the time course of cAMP fluctuations within which cAMP-dependent phosphorylation might be influenced.

We thank Mr. Robert Moats for assistance in performing the Bradford assays, Dr. Lia Faiman for her kind assistance with the biochemical measures, and Drs. Martha Gillette and Jonathan Sweedler for careful reading of the manuscript.

This work was supported by National Institutes of Health (RO1 NS26838) and National Science Foundation (IBN 88-21219).

The derivation of the differential equation follows from that established for solutions in three dimensional electrotonus theory (Purves, 1976, 1977). The following normalized variables are used: B=r_n(k_h/D)^0.5, R=r(k_h/D)^0.5, where r, r_n, D, and k_h have the same meanings as in the text.

The general form of the differential equation describing movement of potential in a three-dimensional syncytium (Purves, 1976, 1977) as recast in terms of movement of ions. Since at steady state dC/dt = 0, the differential equation simplifies to:

The general solution to Eq. A1 is:

The following boundary conditions apply to Eq. A2:

(A3)the initial conditions are: 0 ≤ R ≤ B

(A4)impermeable cellular membrane at B: R ≤ B

source at radius R= B:

Solving the subsidiary equations with the boundary conditions yields:

The restriction X₁ = −X₂ must also be applied to Eq. A2 to prevent [cAMP] from approaching infinity as R → 0 (Purves, 1976, 1977). Using the relationship from Eq. A6 and the restriction that X₁ = −X₂, Eq. A2 simplifies to Eq. 3.