Introduction

Top Abstract Introduction Methods Results Discussion References

Because of the widespread use of caffeine and related xanthines as constituents of food and beverages and as therapeutic drugs, identification of mechanisms that mediate their pharmacological effects has considerable relevance for drug development and therapeutics. Numerous studies have demonstrated that caffeine and other xanthines can have significant effects on a variety of behavioral measures including schedule-controlled behavior (Glowa and Spealman, 1984; Katz and Goldberg, 1987; Spealman, 1988; Howell, 1993a), drug self-administration (Griffiths et al., 1979; Carroll et al., 1989), delayed matching-to-sample (Hudzik and Wenger, 1993) and repeated acquisition (Evans and Wenger, 1992; Buffalo et al., 1993). Over a range of doses, xanthines have significant behavioral-stimulant effects indicative of CNS stimulation (Spealman, 1988; Coffin and Spealman, 1989; Howell, 1993a; Howell and Byrd, 1993), whereas higher doses can suppress behavioral activity (Katz and Goldberg, 1987; Katz et al., 1988; Spealman, 1988; Howell, 1993a) and disrupt behavioral performances associated with learning and memory (Buffalo et al., 1993; Evans and Wenger, 1992; Hudzik and Wenger, 1993). Xanthines also have pronounced respiratory-stimulant effects that have been used in the treatment of respiratory disorders (Rall, 1990), and experimental evidence indicates that xanthines may act by affecting central mechanisms controlling respiration (Lundberg et al., 1981; Wessberg et al., 1985; Howell et al., 1990). Both caffeine and theophylline increase minute volume and sensitivity to elevated levels of inspired CO₂ (Shannon et al., 1975; Mazzarelli et al., 1986; Olson and Schlitt, 1981; Howell, 1993a, b; Howell et al., 1990). These effects have been attributed to a lowering of the threshold of central chemoreceptors that are sensitive to CO₂ (Davi et al., 1978; Gerhardt et al., 1979; Howell et al., 1990; Howell, 1993a, b) rather than to peripheral effects in the lung (Gerhardt et al., 1979).

Xanthines exhibit a complex pharmacology and have several neurochemical actions that could mediate their behavioral and respiratory effects. However, two primary mechanisms involving the cyclic nucleotide system have been implicated as the bases for the effects of xanthines in the CNS. A variety of xanthines bind to specific adenosine recognition sites in vitro, and there is a close correspondence between their potency in stimulating locomotor activity in rodents and their ability to block adenosine receptors in vitro (Bruns et al., 1980; Snyder et al., 1981; Katims et al., 1983). Moreover, metabolically stable analogs of adenosine are behavioral depressants, and their effects are attenuated by xanthines in a manner that indicates a competitive pharmacological antagonism (Logan and Carney, 1984; Goldberg et al. 1985; Spealman, 1988; Spealman and Coffin, 1988; Howell, 1993a; Howell and Byrd, 1993). Accordingly, the behavioral-stimulant effects of xanthines have been linked repeatedly to their capacity to block the actions of endogenous adenosine (Fredholm, 1980; Snyder et al., 1981; Daly, 1982; Howell, et al., 1997). In addition to their adenosine-antagonist effects, xanthines inhibit cyclic nucleotide PDEs responsible for the hydrolytic inactivation of cyclic AMP and cyclic GMP (De Gubareff and Sleator, 1965; Daly, 1977, 1982), sometimes at concentrations similar to those capable of blocking adenosine receptors (Smellie et al., 1979; Choi et al., 1988). Evidence indicates that the respiratory-stimulant effects of xanthines are linked more closely to PDE inhibition than to antagonism of adenosine receptors (Howell et al., 1990, 1997; Howell, 1993a). For example, type IV-selective PDE inhibitors have pronounced respiratory-stimulant effects similar to those of xanthines (Howell et al., 1990; Howell, 1993a), and the potencies of several xanthines as respiratory stimulants correspond with their potencies as PDE inhibitors (Howell et al., 1997). Collectively, these findings suggest that different neurochemical substrates involving the cyclic nucleotide system may mediate the effects of xanthines on respiration and behavior.

The study of pharmacological tolerance during chronic drug administration has proven to be an effective approach to characterize drug mechanism of action. Direct comparisons between the rate of development and loss of tolerance, and altered sensitivity to drugs with selective neurochemical actions, can provide a basis for comparing mechanisms that mediate the respiratory and behavioral effects of xanthines. Human and animal studies have reported the development of tolerance to the CNS effects of caffeine that include behavioral effects (Holtzman, 1983; Finn and Holtzman, 1986, 1987, 1988; Holtzman et al., 1991; Evans and Griffiths, 1992) and humoral and cardiovascular effects (Robertson et al., 1981; Ammon et al., 1983). However, the mechanisms involved in caffeine tolerance are poorly understood, and specific neurochemical substrates have not been identified. Therefore, the present study established a nonhuman primate model of caffeine tolerance to investigate neurochemical mechanisms that mediate caffeine-induced changes in respiration and behavior. The effects of caffeine were compared with those of theophylline, a xanthine with nonselective PDE inhibitory effects and the primary metabolite of caffeine in macaque monkeys (Berthou et al., 1992), rolipram and Ro 20-1724, two type IV (cAMP-specific, cGMP-insensitive)-selective PDE inhibitors and cocaine, a nonxanthine stimulant. In addition, caffeine pharmacokinetics were assessed before and after chronic administration of caffeine.

	Methods

Top Abstract Introduction Methods Results Discussion References

Subjects

Five adult male (CF-67, RBL, RJF, RLM and RQN) and seven adult female (CF-6, CF-65, CF-61, RBS, RPZ, RQC and RYL) rhesus monkeys (Macaca mulatta) weighing between 6.8 and 14.8 kg were studied. Between experimental sessions, subjects lived in individual home cages and were provided daily access to food (commercially available monkey chow, fresh fruit and vegetables) and unrestricted access to water. Weight restrictions were not imposed. Subjects CF-65, CF-67, RBL, RJF, RLM and RQN had been studied previously under the general procedures described below and had received drugs (Howell, 1993a,b, 1995).

Apparatus

Respiration experiments. During experimental sessions, the subjects were seated in a primate chair (Primate Products, Redwood City, CA) and enclosed within a ventilated, sound-attenuating chamber. Ventilation was monitored continuously with a pressure-displacement head plethysmograph, as described previously (Howell et al., 1988). A sealed, rectangular helmet (9.5 × 15 × 11 cm) made of 1/16-inch Lexan with a hole for the neck was placed over the subject's head and served as a pressure-displacement plethysmograph. A continuous flow (10 l/min) of gas, monitored by a flowmeter (model PRO34-FMO34, Cole-Parmer Instrument Co., Chicago, IL), entered the helmet through a port in front of the subject's face and was extracted by a vacuum pump through a port behind the subject's head. A second flowmeter monitored the flow (10 l/min) of gas extracted by the pump. Changes in flow within the plethysmograph resulting from changes in ventilation were measured by a pressure transducer (model MP-15, Micron Instruments, Los Angeles, CA) connected to a polygraph (model 7D, Grass Instruments Co., Quincy, MA). A polygraph integrator (model 7P10F, Grass Instruments Co.) converted the flow signal to a volume measure. A microcomputer interfaced to the polygraph and to the polygraph integrator analyzed and stored data. Respiratory frequency (f) was determined directly by counting changes from positive flow to negative flow. Minute ventilation (V_E) was determined by integration of the pressure-compensated plethysmograph signal after calibrated offset for bias flow, and tidal volume (V_T) was calculated as the quotient of V_E and f, i.e. V_T = V_E/f. Continuous white noise and an exhaust fan masked extraneous sounds during all sessions.

Behavioral experiments. During experimental sessions, subjects were seated in a primate chair and enclosed within a ventilated, sound-attenuating chamber. A response panel mounted on the front of the chair was equipped with a response lever (BRS/LVE, Inc., Laurel, MD, model PRL-001) and a pair of 5-W a.c. red and white lights. The subject's tail was held motionless in a small cuff, and two brass plates rested on a shaved portion near the end. Electrode paste (EGK Sol) minimized changes in impedance between the tail and the brass plates when a 10-mA electric stimulus of 200-msec duration was delivered. A microcomputer controlled experimental events and recorded and stored data on diskettes. Continuous white noise and an exhaust fan masked extraneous sounds during all sessions.

Pharmacokinetic experiments. Drug plasma levels were analyzed by a reverse-phase HPLC analytical system (single pump ternary HPLC system 3004, Isco, Inc., Lincoln, NE) with a variable wavelength ultraviolet absorbance detector (model V4) and a C-18, Spherisorb ODS-2 (5.0 µm, 4.6 × 250 mm) analytical column (Isco, Inc.). A microcomputer and Chemresearch control system software package (Isco, Inc.) commanded the HPLC system, monitored the eluate for absorbance and determined peak areas.

Procedure

Respiration experiments. Subjects CF-65, CF-67 and RQN were adapted to the experimental conditions and stable patterns of ventilation were established before the present studies. Ventilation was recorded in isolated, undisturbed subjects breathing air for an initial 25 to 30 min. Subsequently, subjects were exposed to hypercapnic conditions during which saline and drug injections were given i.m. into the thigh or calf muscle 30 min apart with the following exposure sequence: 10 min of air; 5 min of 3% CO₂ in air; 5 min of 4% CO₂ in air; 5 min of 5% CO₂ in air; and 5 min of air. Injections were given at the start of the 10-min exposure to air to allow for absorption and distribution of the drug before subjects were exposed to elevated levels of CO₂ mixed in air. Generally, two doses of a drug were studied during daily sessions lasting 2 to 3 hr, and drug experiments were conducted no more frequently than twice per week in each subject.

Behavioral experiments. Subjects RBL, RJF and RLM were trained to press a response lever under an FI 300-sec schedule of stimulus termination with a 10-sec limited hold. A red light illuminated the experimental chamber during the interval and the limited hold. If the subject responded during the limited hold after the 300-sec interval elapsed, a white light was illuminated for 2 sec, followed by a 60-sec time-out period during which the chamber was darkened and responses had no scheduled consequences. In the absence of a response during the limited hold, an electric stimulus was delivered, followed by a 60-sec time-out. Daily experimental sessions comprised five sequential components of the FI schedule, and each component included an extended (10-min) time-out period, followed by three consecutive FIs. The total session time was approximately 140 min, and sessions were conducted 5 days per week. The effects of drugs on behavior were determined by a cumulative-dosing procedure similar to that described by Kelleher and Goldberg (1979) and Wenger (1980). A complete dose-effect curve was established during a single session for each drug by injecting (i.m.) sequentially higher doses in the thigh or calf muscle. Saline and drug injections were administered at the start of the extended (10-min) time-out periods. Typically, drug experiments were conducted on Tuesdays and Fridays, and saline (control) was administered on Thursdays. The effects of a full range of doses of each drug were determined at least twice in each subject. To preclude the presentation of excessive stimulus presentations due to drug-induced impairment of responding, the 10-mA stimulus source was disconnected on days when cumulative-dosing procedures were used. The subjects seldom received a stimulus presentation when the stimulus unit was connected, and responding was well maintained when the stimulus unit was disconnected.

Pharmacokinetic experiments. Subjects CF-6, CF-61, RBS, RPZ, RQC and RYL were trained to extend a leg through the front of the home cage for repeated blood withdrawals. After administration (i.m.) of 10.0 mg/kg caffeine, 1.0 ml of blood was withdrawn from the saphenous vein at 0.5, 1.0, 2.0, 6.0 and 24.0 hr postinjection. Caffeine and two primary metabolites, theophylline and paraxanthine, were extracted and assayed by a procedure described previously (Howell, 1995). Whole blood was collected in vacutainer tubes and centrifuged at 3,000 rpm for 10 min. Serum was then aspirated into collection tubes, and 40 µl of distilled water and 10 µl of 60% perchloric acid were added to the sample. After the sample was vortexed and placed in the freezer for 10 min, an additional 350 µl of distilled water was added, and the sample was centrifuged again. The supernatant was transferred to microfiltration tubes (PN MF-5500, Bioanalytical Systems, Inc., W. Lafayette, IN) containing 0.45 µm membranes (PN MF-5655) and centrifuged a third time. The filtered product was then injected into the HPLC system with a mobile phase comprising 65% 50 mM sodium phosphate buffer (pH = 2.0) and 35% methanol at a flow rate of 1.0 ml/min. The eluate was monitored for absorbance at 280 nm, and the limit of detection for caffeine and its metabolites was 200 ng/ml serum. Drug-free samples of blood were spiked with known quantities of drug (1.0-50.0 µg/ml) and carried through the extraction procedure to calibrate the process and to determine the percentage yield of recovery.

Drug administration. On multiple occasions, caffeine (10.0 mg/kg) was administered i.m. on a chronic, daily basis with each occasion separated by at least 4 weeks of drug abstinence. During the conduct of experiments, the daily dose of caffeine was administered 10 min presession. To assess caffeine tolerance and drug cross-tolerance, the acute effects of caffeine and several other drugs were determined before and during chronic caffeine administration. When the acute effects of a drug were determined in caffeine-tolerant subjects, the daily dose of caffeine was not administered on that day.

Data Analysis

Results were calculated for individual subjects and combined to derive the mean for the group. In respiration studies, the last 3 min of each 5-min exposure to CO₂ were used for data analysis to allow time for ventilation to reach a steady state. In behavioral experiments, response rates were computed separately for each FI component by dividing the total number of responses in a component by the total time the red light was present. Mean control rates were determined for each monkey by averaging response rates for all saline control sessions. The effects of drugs on responding were expressed as a percent of control rate obtained when saline was administered. In pharmacokinetic experiments, caffeine half-life (t_1/2) was calculated by least-squares linear-regression analysis of ln(C_t/C_peak) as a function of time where C_t = concentration at time t, C_peak = peak concentration and ln(C_t/C_peak = (0.693/t_1/2)t. A repeated-measures analysis of variance with Tukey's post hoc multiple comparisons was used to determine the statistical significance of drug treatment conditions, and statistical significance was accepted at the 95% level of confidence (P < .05). In some figures, the S.E.M. or 95% confidence limits were presented only for the control data to avoid figures that appeared cluttered and difficult to interpret.

Drugs

The drugs studied were caffeine and theophylline (Sigma Chemical Co., St. Louis, MO), cocaine hydrochloride (National Institute on Drug Abuse, Rockville, MD), rolipram (Berlex Laboratories, Inc., Cedar Knolls, NJ) and Ro 20-1724 (Research Biochemicals, Inc., Natick, MA). Caffeine and theophylline were dissolved in 0.9% saline containing sodium benzoate, and doses are expressed in terms of the free base of the drugs. Ro 20-1724 was dissolved in 45% (w/v) aqueous 2-hydroxypropyl--cyclodextrin. Cocaine and rolipram were dissolved in 0.9% saline. Injection volumes were 0.5 to 3.0 ml.

	Results

Top Abstract Introduction Methods Results Discussion References

Respiration Experiments

During control conditions when subjects breathed air alone, f averaged 19.0 ± 1.5 breaths/min, V_T averaged 78.9 ± 9.8 ml and V_E averaged 1.5 ± .2 l/min for the group of three monkeys. Exposure to elevated levels of CO₂ in air produced marked increases in ventilation above levels recorded during exposure to air alone (fig. 1). There was a concentration-dependent increase in f, V_T and V_E as inspired CO₂ increased to a maximum of 5%. During the first 2 to 3 min of subsequent exposure to air, ventilation decreased and returned to prior values. Polygraph tracings also illustrate caffeine-induced changes in ventilation (fig. 1). On the first day of chronic, daily administration of 10.0 mg/kg caffeine, there was a marked increase in ventilation during exposure both to air alone and to elevated levels of CO₂ in air. However, caffeine had little effect on ventilation by day 8 of chronic administration.

View larger version (34K): [in this window] [in a new window]   Fig. 1.   Polygraph recordings of ventilation in subject CF-65. Ventilation during exposure to air alone (normocapnia) is shown in the top panel. Ventilation during exposure to elevated levels of CO2 mixed in air (hypercapnia) and the effects of chronic, daily administration of 10.0 mg/kg caffeine (days 1 and 8) on the ventilatory response to hypercapnia are shown in the bottom panels. Arrows indicate the beginning of 5-min exposures to 3%, 4% and 5% CO2 mixed in air and subsequent exposure to air alone. Abscissae, time and concentration of inspired gas; ordinates, continuous flow. See Howell et al. (1988) for a more detailed presentation regarding the time-course effects of CO2 exposure.

In figure 2, the data are summarized for the group of three monkeys, and f and V_E are shown as a function of CO₂ concentration during control conditions and after caffeine administration. On the first day of chronic administration, the effects of caffeine were similar to those obtained previously in acute administration experiments. Caffeine produced significant increases in f and V_E during exposure to air alone and during exposure to 3%, 4% and 5% CO₂ in air (fig. 2, top). Caffeine-induced increases in V_E were caused primarily by increases in f with little change in V_T. However, by day 5 of chronic administration, the effects of caffeine were less pronounced (data not shown), and by day 8, neither f nor V_E differed significantly from nondrug, baseline conditions. When chronic administration was terminated and the acute effects of caffeine were redetermined, sensitivity returned to levels obtained before chronic administration within 9 days, and caffeine produced significant increases in f and V_E (fig. 2, bottom). Moreover, control ventilation was characteristic of prechronic conditions, providing no evidence of drug withdrawal.

View larger version (24K): [in this window] [in a new window]   Fig. 2.   Development of tolerance to the respiratory effects of caffeine during chronic, daily administration of 10.0 mg/kg caffeine for 8 consecutive days (top) and loss of tolerance when chronic administration was terminated (bottom). Subjects were administered caffeine daily during exposure to air alone (normocapnia) and during exposure to 3%, 4% and 5% CO2 mixed in air (hypercapnia). Each point is the mean (± S.E.M.) of three subjects. Open circles represent control values obtained after saline administration before chronic caffeine administration. Abscissae, concentration of CO2 mixed in air; ordinates, respiratory frequency (breaths per minute) and minute volume (liters per minute).

In subsequent experiments, 10.0 mg/kg caffeine was administered daily until complete tolerance was evident, and on day 9, the acute effects of 30.0 mg/kg caffeine were determined. The higher dose of caffeine produced significant increases in f and V_E, but the effects were similar to those obtained with 10.0 mg/kg in nontolerant subjects (fig. 3, top). In another series of experiments, the effects of rolipram (0.03 mg/kg) were compared with those of caffeine in nontolerant and caffeine-tolerant subjects. The acute effects of rolipram were very similar to those of caffeine in nontolerant subjects. Rolipram produced significant increases in f and V_E during exposure to air alone and during exposure to 3%, 4% and 5% CO₂ in air (fig. 3, bottom). However, compared with nontolerant subjects, the acute effects of rolipram were less pronounced during chronic administration of caffeine.

View larger version (27K): [in this window] [in a new window]   Fig. 3.   Respiratory effects of 30.0 mg/kg caffeine (top) or 0.03 mg/kg rolipram (bottom) in nontolerant and caffeine-tolerant subjects. A dose of 30.0 mg/kg caffeine in tolerant subjects had effects that were similar to 10.0 mg/kg in nontolerant subjects. Also, caffeine-tolerant subjects were cross-tolerant to the selective PDE inhibitor, rolipram. Abscissae, concentration of CO2 mixed in air. Open circles represent control values obtained after saline administration before chronic caffeine administration. Ordinates, respiratory frequency (breaths per minute) and minute volume (liters per minute). Otherwise, as in figure 2.

Additional experiments investigated the respiratory effects of theophylline and Ro 20-1724 administered as cumulative doses in nontolerant and caffeine-tolerant subjects. Like rolipram, the acute effects of theophylline (10.0 and 30.0 mg/kg) and Ro 20-1724 (0.1 and 0.3 mg/kg) were very similar to those of caffeine in nontolerant subjects. Both drugs produced significant, dose-related increases in f during exposure to air alone and during exposure to 3%, 4% and 5% CO₂ in air (fig. 4). Drug-induced increases in V_E (data not shown) were caused primarily by increases in f, and closely paralleled drug effects on f shown in figure 4. Compared with nontolerant subjects, the acute effects of theophylline and Ro 20-1724 were less pronounced during chronic administration of caffeine. For example, effective doses of theophylline (10.0 mg/kg) and Ro 20-1724 (0.1 mg/kg) in nontolerant subjects had no significant effects on ventilation in caffeine-tolerant subjects.

View larger version (28K): [in this window] [in a new window]   Fig. 4.   Respiratory effects of 10.0 and 30.0 mg/kg theophylline (top) or 0.1 and 0.3 mg/kg Ro 20-1724 (bottom) administered as cumulative doses in nontolerant and caffeine-tolerant subjects. Open circles represent control values obtained after saline administration before chronic caffeine administration. Note that caffeine-tolerant subjects were cross-tolerant to the caffeine metabolite, theophylline, and to the selective PDE inhibitor, Ro 20-1724. Abscissae, concentration of CO2 mixed in air; ordinates, respiratory frequency (breaths per minute). Otherwise, as in figure 2.

A final series of experiments investigated potential changes in baseline sensitivity to CO₂ during chronic administration of caffeine. To preclude acute drug effects during CO₂ determinations, the daily dose of caffeine was administered 10 min postsession. Chronic administration of 10.0 mg/kg caffeine for 8 consecutive days had no significant effect on sensitivity to CO₂; both nontolerant and caffeine-tolerant subjects exhibited similar CO₂-induced increases in f and V_E (fig. 5).

View larger version (13K): [in this window] [in a new window]   Fig. 5.   Ventilatory response to hypercapnia in nontolerant and caffeine-tolerant subjects. The chronic, daily dose of caffeine was administered postsession to preclude acute drug effects during CO2 determinations. Note that chronic administration of caffeine had no significant effect on the ventilatory response to hypercapnia. Otherwise, as in figure 2.

Behavioral Experiments

Responding during the FI 300-sec schedule was characteristic of performance maintained under FI schedules of stimulus termination (Morse and Kelleher, 1966). Little or no responding occurred early in the interval, and the response rate increased as the interval elapsed. Mean response rates during control sessions were 0.89 ± 0.34, 0.40 ± 0.14 and 0.37 ± 0.10 response/sec for subjects RJF, RBL and RLM, respectively. Caffeine (10.0 mg/kg) produced significant increases in response rate to more than 200% of control values for the group of three monkeys during the first day of chronic administration (fig. 6). However, the behavioral-stimulant effects of caffeine gradually diminished during 5 consecutive days, and by day 5, caffeine had no significant effect on FI responding. When chronic administration of caffeine was terminated on day 20, there was no obvious behavioral disruption characteristic of drug withdrawal.

View larger version (17K): [in this window] [in a new window]   Fig. 6.   Development of tolerance to the behavioral effects of caffeine during chronic, daily administration of 10.0 mg/kg caffeine for 19 consecutive days. Each point is the mean (± S.E.M.) of three subjects. Horizontal dashed lines indicate 95% confidence intervals for saline (control) sessions. Data points are missing for weekends (days 6, 7, 13 and 14) when subjects were not tested, and for days (9, 12 and 19) when cumulative-dosing procedures were used for acute dose-effect determinations. Asterisks indicate a significant (P < .05) difference from control rate. Abscissa, consecutive days; ordinate, mean (± S.E.M.) response rate expressed as a percent of control rate obtained when saline (S) was administered.

In subsequent experiments, administration of cumulative doses of caffeine (1.0-30.0 mg/kg) in nontolerant subjects had modest behavioral-stimulant effects (fig. 7) that were less pronounced than those obtained when a single dose of caffeine (10.0 mg/kg) was administered 5 min presession (see fig. 6). A maximum rate increase of approximately 130% of control rate occurred after a dose of 3.0 mg/kg, and the highest dose (30.0 mg/kg) significantly decreased response rate below control values. When subjects received chronic, daily administration of 10.0 mg/kg caffeine for 2 consecutive weeks, there was a significant main effect of chronic treatment during the second week, shown as a downward displacement of the caffeine dose-effect curve (fig. 7, top). However, caffeine-induced suppression of responding after a dose of 30.0 mg/kg was relatively unchanged. When chronic administration was terminated and the acute effects of caffeine were redetermined 7 days later, modest rate-increasing effects were obtained, and the caffeine dose-effect curve did not differ significantly from prechronic conditions.

View larger version (18K): [in this window] [in a new window]   Fig. 7.   Development of tolerance to the behavioral effects of caffeine during chronic, daily administration of 10.0 mg/kg caffeine for 2 consecutive weeks (top) and loss of tolerance when chronic administration was terminated (bottom). Each point is the mean (± S.E.M.) of three subjects. Cumulative-dosing procedures were used for acute dose-effect determinations. Abscissae, caffeine dose, log scale; ordinate, mean (± S.E.M.) response rate expressed as a percent of control rate obtained when saline was administered.

In another series of experiments, the behavioral effects of theophylline (1.0-30.0 mg/kg), rolipram (0.01-0.3 mg/kg) and cocaine (0.03-1.0 mg/kg) were compared with those of caffeine in nontolerant and caffeine-tolerant subjects. Cumulative doses of theophylline had effects that were very similar to those of caffeine in nontolerant subjects (fig. 8, top). A maximum rate increase of approximately 125% of control rate occurred after a dose of 3.0 mg/kg, and the highest dose (30.0 mg/kg) significantly decreased response rate below control values. In contrast, rolipram only decreased response rate over a wide range of doses (fig. 8, center). Cocaine produced dose-related increases in FI responding that were much more pronounced than those of caffeine, and the highest dose (1.0 mg/kg) decreased response rate below the maximum rate (fig. 8, bottom). When subjects received chronic, daily administration of 10.0 mg/kg caffeine for 2 consecutive weeks and the acute effects of theophylline were redetermined during the second week, there was no significant main effect of chronic treatment. However, no dose of theophylline increased FI responding above control rates during chronic caffeine administration. In separate experiments, there was no significant change in sensitivity to the behavioral effects of rolipram and cocaine during the second and third week of chronic caffeine administration, respectively. Although the maximum rate-increasing effect of cocaine was partially attenuated during chronic caffeine administration, pronounced rate-increasing effects were obtained in all subjects, and the cocaine dose-effect curve did not differ significantly from prechronic conditions. Hence, there was some indication of cross-tolerance to theophylline in caffeine-tolerant subjects, but there was little evidence of cross-tolerance to rolipram or cocaine.

View larger version (19K): [in this window] [in a new window]   Fig. 8.   Behavioral effects of theophylline (top), rolipram (center) and cocaine (bottom) before and during chronic, daily administration of 10.0 mg/kg caffeine. Each point is the mean (± S.E.M.) of three subjects. Cumulative-dosing procedures were used for acute dose-effect determinations. Abscissae, drug dose, log scale; ordinates, mean (± S.E.M.) response rate expressed as a percent of control rate obtained when saline was administered.

Pharmacokinetic Experiments

Standard curves for caffeine and its metabolites were linearly related to peak areas over a range of 1.0 to 50.0 µg/ml. Chromatograms were completed in less than 10 min with reliable separation of peak retention times for caffeine, theophylline and paraxanthine at approximately 7.9, 5.3 and 4.9 min, respectively. The percent yield obtained during extractions of standards varied little among xanthines and ranged from 86.9% to 95.6% with a mean (± S.E.M.) value of 92.4% ± 1.7%.

Before chronic administration, 10.0 mg/kg caffeine resulted in a peak plasma concentration (± S.E.M.) of 8.6 ± 0.6 µg/ml at 30 min postinjection for the group of six monkeys (fig. 9). The calculated half-life of caffeine was 10.8 ± 0.6 hr, and as caffeine plasma levels decreased, there was a corresponding increase in theophylline levels. No significant amount of paraxanthine was detected. At 24 hr postinjection, only small amounts of caffeine (<1.0 µg/ml) were detected, whereas significant amounts of theophylline (3.9 µg/ml) were still present. When caffeine pharmacokinetics were redetermined during the second week of chronic, daily administration of 10.0 mg/kg caffeine, neither the peak plasma concentration (8.0 ± 0.2 µg/ml) nor the calculated half-life (10.3 ± 0.9 hr) of caffeine differed significantly from prechronic conditions (fig. 9). However, theophylline levels were significantly greater throughout the 24-hr postinjection period because of accumulation of theophylline from prior daily injections of caffeine. The results indicate that chronic caffeine administration had little effect on caffeine metabolism or clearance.

View larger version (20K): [in this window] [in a new window]   Fig. 9.   Mean (± S.E.M.) plasma levels of caffeine and its primary metabolite, theophylline, after administration of 10.0 mg/kg caffeine in a group of six monkeys. Caffeine pharmacokinetics were determined before and after chronic administration of 10.0 mg/kg caffeine for 10 consecutive days. Note that theophylline levels were significantly greater during chronic caffeine administration, but mean plasma levels and plasma half-lives of caffeine were similar to prechronic conditions.

	Discussion

Top Abstract Introduction Methods Results Discussion References

The present study established a nonhuman primate model of caffeine tolerance to investigate neurochemical mechanisms that mediate the physiological and behavioral effects of caffeine. Chronic daily administration of caffeine resulted in tolerance to its respiratory- and behavioral-stimulant effects that developed gradually over several days, and was reversible when chronic administration was terminated. Caffeine tolerance appeared to be pharmacodynamic because no significant changes in caffeine metabolism or clearance were evident in caffeine-tolerant subjects. Although the time course for the development and loss of tolerance was similar for the respiratory and behavioral effects of caffeine, important characteristics of caffeine tolerance differed for respiration and behavior. Tolerance to the respiratory-stimulant effects of caffeine was surmountable and extended to the type IV PDE inhibitors, rolipram and Ro 20-1724. A high dose of caffeine produced significant increases in ventilation in caffeine-tolerant subjects, and the acute effects of rolipram and Ro 20-1724 were less pronounced during chronic caffeine administration. In contrast, tolerance to the behavioral-stimulant effects of caffeine was insurmountable and did not extend to rolipram. No dose of caffeine increased responding above control rates in caffeine-tolerant subjects, and the acute effects of rolipram were unchanged during chronic caffeine administration. The results indicate that different neurochemical mechanisms mediate the effects of caffeine on respiration and behavior, and that inhibition of type IV PDE plays a prominent role in caffeine-induced respiratory stimulation.

The pharmacological profile of caffeine in the present study is consistent with previous studies reporting the involvement of type IV PDE in the acute respiratory-stimulant effects of xanthines (Howell et al., 1990; Howell, 1993a). Studies in rhesus monkeys have compared the effects of caffeine, theophylline, 8-phenyltheophylline, 8-cyclopentyltheophylline, 3-isobutyl-1-methylxanthine and enprofylline, as well as the nonxanthine adenosine antagonist, CGS 15943, and the selective PDE inhibitors, rolipram and Ro 20-1724. All drugs except 8-phenyltheophylline and CGS 15943, potent adenosine antagonists lacking PDE inhibitory effects, were found to have prominent respiratory-stimulant effects. Enprofylline and rolipram, both of which have prominent PDE-inhibitory effects but low affinity for adenosine receptors, had respiratory effects similar to those of caffeine. When the effects of prototypic xanthines were compared with nonxanthines that inhibit different molecular forms of PDE, only rolipram and Ro 20-1724, which selectively block the type IV (cAMP-specific, cGMP-insensitive) PDE, had respiratory effects in common with xanthines, and the drug potencies as respiratory stimulants correlated with their potencies as PDE inhibitors (Howell et al., 1997). The results obtained in the present study further support a prominent role of type IV PDE in the respiratory effects of xanthines. Theophylline, rolipram and Ro 20-1724 had acute respiratory effects that were similar to those of caffeine in nontolerant and caffeine-tolerant subjects. Moreover, the demonstration of cross-tolerance to theophylline and the type IV PDE inhibitors in caffeine-tolerant subjects provides convincing evidence that these drugs share a common mechanism of action.

Caffeine had pronounced effects on ventilation during conditions of normocapnia, and produced dose-dependent increases in f and V_E during exposure to elevated levels of CO₂ mixed in air such that the CO₂ response curve was shifted upward in a parallel manner. The effects observed during hypercapnia are consistent with previous studies reporting that xanthines act by affecting central mechanisms controlling respiration (Howell et al., 1990; Lundberg et al., 1981; Trippenbach et al., 1980; Wessberg et al., 1985). For example, caffeine increases the ventilatory response to CO₂ in cats at levels of CO₂ equal to or below the apneic range (Mazzarelli et al., 1986), and theophylline increases the ventilatory response to CO₂ in dogs during hypercapnia (Olson and Schlitt, 1981). Accordingly, a series of experiments investigated potential changes in baseline sensitivity to CO₂ during chronic administration of caffeine to determine whether physiological adaptation resulted from repeated exposure to hypercapnia. When the chronic daily dose of caffeine was administered postsession to preclude acute drug effects during CO₂ determinations, caffeine-tolerant subjects exhibited CO₂-induced increases in ventilation that were similar to those obtained in nontolerant subjects. The latter results indicate that caffeine tolerance was pharmacological and did not result from changes in baseline responsiveness to CO₂.

The pharmacokinetics of caffeine and its primary metabolites have been well characterized in humans and in a variety of laboratory animals, but the direct involvement of pharmacokinetic factors in caffeine tolerance has not been demonstrated. The major route of caffeine metabolism is through liver biotransformation to several dimethylxanthine metabolites including theophylline, paraxanthine and theobromine (Bonati et al., 1985; Lelo et al., 1986), and significant species differences in xanthine metabolism are related to multiple isoforms of the P450 enzyme (Berthou et al., 1992). For example, N-3 demethylation to paraxanthine is the major metabolic pathway in humans, whereas N-7 demethylation to theophylline is predominant in cynomolgus monkeys (Berthou et al., 1992) and rhesus monkeys (Howell, 1995). In the present study, the calculated half-life of caffeine in nontolerant subjects was approximately 11 hr, and as caffeine plasma levels decreased, there was a corresponding increase in theophylline levels. When caffeine pharmacokinetics were redetermined in caffeine-tolerant subjects, neither the peak plasma concentration nor the calculated half-life of caffeine differed from nontolerant subjects. The latter results indicate that chronic administration of caffeine had little effect on caffeine metabolism or clearance and that caffeine tolerance resulted from pharmacodynamic rather than pharmacokinetic factors. Moreover, the long duration of action of caffeine in the rhesus monkey may explain how a single daily dose of caffeine could result in caffeine tolerance during the course of a week. The accumulation of theophylline, an active metabolite of caffeine, during chronic caffeine administration suggests that caffeine tolerance was actually underestimated. Similarly, the gradual decline in caffeine blood levels with a corresponding increase in theophylline may account for the absence of overt withdrawal signs in the present study.

In contrast to the respiratory-stimulant effects of caffeine and related xanthines, the behavioral-stimulant effects have been linked more closely to adenosine receptor antagonism than to PDE inhibition (Bruns et al., 1980; Katims et al., 1983; Snyder et al., 1981). In previous studies conducted in nonhuman primates, only xanthines with prominent adenosine-antagonist actions had significant behavioral-stimulant effects, and there was a close correspondence between the drug potencies for increasing response rate and for antagonizing the behavioral-suppressant effects of the adenosine receptor agonist, NECA (Spealman, 1988; Howell, 1993a; Howell and Byrd, 1993; Coffin and Spealman, 1989). Selective PDE inhibitors that lacked adenosine-antagonist effects either suppressed behavior or had no behavioral effect (Howell, 1993a). Consistent with these findings, rolipram only decreased response rate in the present study, and subjects tolerant to the behavioral-stimulant effects of caffeine were not cross-tolerant to rolipram. The pharmacological specificity of caffeine tolerance was exemplified further by the lack of cross-tolerance to the nonxanthine psychomotor stimulant, cocaine. In contrast to rolipram, acute administration of theophylline, a xanthine with adenosine-antagonist effects, had modest behavioral-stimulant effects similar to those of caffeine. Although there was no significant main effect of chronic caffeine treatment on the acute effects of theophylline, no dose of theophylline increased responding above control rates during chronic caffeine administration. Hence, there was some indication of cross-tolerance to theophylline in caffeine-tolerant subjects. However, the role of adenosine antagonism as a mechanism underlying tolerance to caffeine-induced stimulation of behavior remains undefined. Some investigators have reported an increase in the number of adenosine binding sites in brain during chronic administration of caffeine (Ahlijanian and Takemori, 1986; Hawkins et al., 1988), whereas others have found no appreciable changes in the potency of caffeine as an adenosine antagonist in caffeine-tolerant animals (Holtzman et al., 1991; Katz and Goldberg, 1987).

In summary, chronic administration of caffeine resulted in tolerance to its respiratory- and behavioral-stimulant effects which developed gradually and was reversible. Caffeine pharmacokinetics were unchanged during chronic administration, which indicated that caffeine tolerance was pharmacodynamic. Tolerance to the respiratory-stimulant effects of caffeine was surmountable and extended to the type IV PDE inhibitors, rolipram and Ro 20-1724, whereas tolerance to the behavioral-stimulant effects was insurmountable and did not extend to rolipram. The results suggest that different neurochemical mechanisms mediate the effects of caffeine on respiration and behavior and that inhibition of type IV PDE plays a prominent role in caffeine-induced respiratory stimulation.