Introduction

Abstract Introduction Materials & Methods Results Discussion References

Cigarette smoking is the leading cause of preventable death in the United States (McGinnis and Foege, 1993). Recent advances in behavioral and pharmacologic treatment approaches have improved a smoker's chances of quitting (Fiore et al., 1994; Henningfield, 1995; Hjalmarson et al., 1994). Nevertheless, the vast majority of those who try to quit will fail.

A pivotal role for nicotine in maintaining cigarette smoking is well established. Smokers tightly regulate their nicotine intake. Subjects who are switched from their usual brand to lower-nicotine cigarettes alter their smoking behavior to increase their nicotine yield per cigarette (Benowitz et al., 1986). Conversely, supplementation of nicotine intake by i.v. infusion reduces cigarette smoking (Lucchesi et al., 1967). Cessation of nicotine intake produces a withdrawal syndrome, and replacement of nicotine reduces its severity (Hughes et al., 1984). Thus an intervention aimed at reducing nicotine distribution to target tissues such as brain might provide a strategy for promoting cessation of smoking.

The possibility of using drug-specific antibodies to reduce the effects of drugs of abuse was studied more than 20 years ago by Bonese et al. (1974), who actively immunized one monkey against heroin after it had been trained to self-administer both heroin and cocaine. Immunization reduced the self-administration of heroin but not of cocaine, a result that demonstrated both its efficacy and its specificity. Similar efficacy was produced by transfusing two nonimmunized monkeys with morphine-specific antiserum from immunized monkeys (Killian et al., 1978). Altered patterns of heroin self-administration lasted more than 3 weeks, the effect of immunization waning in parallel with antibody titers.

More recently, it has been shown that the behavioral effects of cocaine can be modified by active immunization. Immunized rats challenged with a substantial dose of cocaine (15 mg/kg i.p.) showed significantly reduced locomotor activity and lower brain cocaine concentrations than controls (Carrera et al., 1995). In a separate study, self-administration of i.v. cocaine by rats at clinically relevant doses was rapidly suppressed by the passive administration of cocaine-specific antibodies (Fox et al., 1996). This result is particularly interesting because the cocaine dose delivered by each infusion (1 mg/kg = 1 µmol per 300-g rat) greatly exceeded the binding capacity of drug-specific antibody administered (4 mg = 0.05 µmol of binding sites). Presumably, the observed blunting of cocaine distribution to brain was sufficient to alter its behavioral effects.

The general concept of immunization as a means of blocking the effects of drugs of abuse is attractive because of its specificity; as a pharmacokinetic intervention that acts outside the CNS and does not affect neurotransmitters or receptors, it should not have many of the adverse effects associated with other treatments. Nicotine is an even better candidate for this approach than heroin or cocaine, because the dose administered is much lower. The dose of nicotine (molecular weight 162 D) delivered by one cigarette is approximately 1 mg, whereas that of a single psychoactive dose of cocaine (molecular weight 303 D) is about 35 mg, a molar ratio of 1:18 (Benowitz and Jacob, 1984; Pentel et al., 1994). The molar ratio of total daily doses (nicotine, 37 mg; cocaine, 1 g) is about the same. Thus the efficacy of immunization in modifying the behavioral effects of cocaine suggests that it could be even more effective in altering the behavioral effects of nicotine. Nicotine is also a good candidate for this approach because many smokers who are trying to quit are highly motivated and thus are unlikely to try to overcome the effect of antibody by purposely increasing their nicotine intake.

As a first step in evaluating this question, the current study examined the effects of active immunization on the distribution kinetics of nicotine in rats. We hypothesized that immunization would alter nicotine distribution by increasing nicotine binding in plasma and reducing the nicotine concentration in brain. A pharmacokinetically relevant model that mimics the marked arterio-venous nicotine concentration gradient produced by cigarette smoking was used, along with clinically relevant doses of nicotine.

	Materials and Methods

Abstract Introduction Materials & Methods Results Discussion References

Drugs and reagents. (±)-Nicotine sulfate, ()-cotinine, ()-[methyl-³H]-nicotine 70 Ci/mmol, and goat anti-IgG-peroxidase conjugate were obtained from Sigma Chemical Co. (St. Louis, MO). (±)-Nicotine-N-oxide was a gift from Professor John Gorrod. Internal standards for the nicotine/cotinine assay (5-methyl nicotine bis-oxalate and 5-methyl cotinine perchlorate) were a gift from Dr. Peyton Jacob.

Synthesis of immunogen. (fig. 1) Production of CMUNic was accomplished by the synthesis of 6-amino-(±)-nicotine, reaction with ethyl isocyanatoacetate to form 6-(carboxyethylureido)-(±)-nicotine and hydrolysis by lithium hydroxide to CMUNic. All structures were confirmed by nuclear magnetic resonance (360 or 90 MHz), infrared spectra and elemental analysis.

View larger version (12K): [in this window] [in a new window]   Fig. 1.   Structure of the nicotine immunogen CMUNic. Immunogen was bound to carrier protein by carbodiimide linkage at the terminal carboxyl group.

Immunization of rats. Male Holtzman rats weighing 175 to 200 g were immunized with 25 µg of CMUNic-KLH or KLH alone in 0.4 ml of complete Freund's adjuvant injected i.p. At 21 and 35 days, animals were boosted with 25 µg of CMUNic-KLH or KLH alone in 0.4 ml of incomplete Freund's adjuvant injected i.p. Experiments were performed 4 to 12 days after the second booster injection. Nicotine-specific IgG titers were stable over this period of time. At the time of the experiment, animals weighed 285 to 400 g. IgG titers were measured from plasma obtained at the beginning of each experiment.

Characterization of antibodies. We measured serum IgG titers and specificity by ELISA (Keyler et al., 1994), using CMUNic-albumin as the coating antigen to avoid detecting antibodies directed at KLH and using goat anti-rat IgG coupled to peroxidase as the detecting antibody. Specificity was determined by competition with CMUNic, nicotine, acetylcholine (ACh) and the two major nicotine metabolites in the rat, cotinine and nicotine-N-oxide.

Protocol 1: Arterio-venous nicotine concentrations. For studies of arterio-venous nicotine concentrations, eight male Holtzman rats weighing 210 to 270 g were anesthetized with Innovar Vet (droperidol 4 mg/ml and fentanyl 0.08 mg/ml) 1 ml/kg i.m., and PE50 catheters were placed in the left femoral artery and vein and in the right jugular vein. ³H()-nicotine (3 × 10⁶ dpm) in 0.25 ml of 0.9% saline plus unlabeled nicotine 0.03 mg/kg was infused via the jugular vein over 10 sec. Heparinized blood samples of 0.3 ml were removed from the femoral arterial and venous catheters over 15-sec intervals (0-15, 16-30, 31-45, 46-60, 61-75, 76-90 sec), centrifuged immediately and assayed by scintillation counting. Terminal (90-sec) plasma samples from six additional animals were assayed by gas chromatography for cotinine to determine whether any measurable metabolism of nicotine had taken place over this period of time.

Protocol 2: Effects of immunization. For studies of the effects of immunization on nicotine distribution, groups of eight animals immunized with CMUNic-KLH or KLH alone were anesthetized with Innovar Vet, and PE50 catheters were placed in the left femoral and right jugular veins. Nicotine 0.03 mg/kg was administered over 10 sec via the jugular vein catheter, 1.0-ml blood samples were obtained at 10, 20 and 40 min from the femoral vein catheter and plasma was stored at 20°C for measurement of plasma nicotine concentrations. At 40 min, an additional 4 to 5 ml of blood was obtained for measurement of protein binding, animals were sacrificed with KCl and organs were perfused with normal saline by injecting 30 ml into the left ventricle of the heart over 3 min. The brain was rapidly removed and the cortex and cerebellum stored at 20°C for assay of nicotine concentrations.

Protein binding. Protein binding was measured by equilibrium dialysis for 4 h at 37°C using 0.7 ml of plasma, Teflon semi-micro cells, Spectrapor 2 membranes with a molecular weight cutoff of 12 to 14 kD and Sorenson's buffer (0.13 M phosphate, pH 7.4) (Pentel and Keyler, 1987). Plasma pH was measured at the end of each run, and samples were used only if the final pH was 7.30 to 7.45.

Nicotine assay. Nicotine and cotinine concentrations in plasma were measured by gas chromatography with nitrogen-phosphorus detection (Jacob et al., 1981). Sensitivity of the assay is 1 ng/ml nicotine and 5 ng/ml cotinine. Brain samples were digested in 5 volumes of NaOH overnight and homogenized, and internal standards were added. The pH was adjusted to less than 4.0 with 1 M sulfuric acid, and 3 ml of Toluene/1-butanol (70:30) was added. The suspension was mixed and centrifuged and the organic layer removed. The aqueous layer was decanted from the remaining tissue and extracted in the same manner as plasma samples. Nicotine and cotinine recovery from brain extracted in this manner is greater than 90%.

Calculation of nicotine binding capacity in plasma. Nicotine binding capacity in plasma of immunized rats was calculated from the in vivo data as the difference in mean plasma nicotine concentration between CMUNic-KLH and KLH groups at 10 min multiplied by the estimated plasma volume of the rat, 40 ml/kg (Cocchetto and Bjornsson, 1983). Nicotine binding capacity in serum was also calculated from the serum radioimmunoassay data using the method of Muller (1983). The molecular weight of IgG was assumed to be 150 kD.

Statistical methods. Plasma radiolabel and nicotine concentrations were compared between groups by repeated-measures ANOVA and Scheffe's contrast. When group differences were significant, individual time-points were compared using one-tailed unpaired t tests. Differences in brain nicotine concentrations, protein binding of nicotine in plasma and the brain/plasma nicotine concentration ratio between groups were compared using two-tailed unpaired t tests. One animal in the CMUNic group, whose plasma showed no binding activity for nicotine by ELISA (determined after the experiments were run), was excluded from the analysis of the effects of immunization on nicotine distribution.

	Results

Abstract Introduction Materials & Methods Results Discussion References

Antibody characteristics. Among animals immunized with CMUNic-KLH, IgG titers measured by ELISA were all greater than 1:10,000, with the exception of one animal with a titer of 1:2500 that showed inhibition of binding by CMUNic but not by nicotine (this animal was excluded from further analysis). Titers were generally the same after one or two immunogen boosts. ELISA inhibition curves showed that CMUNic had a lower IC₅₀ than nicotine (1.5 × 10⁶ vs. 5.8 × 10⁴, fig. 2). The nicotine metabolites cotinine and nicotine-N-oxide produced very little inhibition at concentrations up to 10² M and no more than the irrelevant control propranolol. ACh also produced no more inhibition than propranolol (data not shown). All animals immunized with KLH alone had titers of less than 1:100.

View larger version (16K): [in this window] [in a new window]   Fig. 2.   Antibody specificity; percent inhibition of antibody binding as determined by ELISA using CMUNic linked to BSA as the coating antigen. CMUNic had a lower IC50 than nicotine, whereas the nicotine metabolites cotinine and nicotine-N-oxide had higher values and showed no more inhibition than the irrelevant control propranolol.

1

6

Arterio-venous nicotine concentrations. Assay of terminal (90-sec) plasma samples by gas chromatography detected no cotinine, so radiolabel concentrations may be taken as representative of the parent compound, nicotine. Arterial radiolabel concentrations exceeded venous concentrations (P < .001), the difference being greatest initially and diminishing rapidly thereafter (fig. 3).

View larger version (17K): [in this window] [in a new window]   Fig. 3.   Mean ± S.D. femoral arterial and venous plasma radiolabel concentrations after administration of 3H-(±) nicotine 0.03 mg/kg as a 10-sec infusion in the right jugular vein (eight rats). Sampling times are plotted as the midpoint of each collection interval. The initial arterial concentrations are substantially higher than venous concentrations (P < .001). Nicotine metabolism over this period of time is negligible, so radiolabel concentrations accurately reflect nicotine concentrations.

Effects of immunization on nicotine distribution. Mean plasma nicotine concentrations were 4 to 6-fold greater in rats immunized with CMUNic-KLH than in controls immunized with KLH alone (P < .001, fig. 4). The higher nicotine concentration in the CMUNic-KLH group was apparent at the first sampling time (10 min). Plasma nicotine concentrations decreased modestly over the next 30 min in both groups, but the difference between groups was sustained. The mean brain nicotine concentration was 13.8% lower in the CMUNic-KLH group than in the KLH group (37.9 ± 4.5 vs. 44.0 ± 8.4 ng/g), but this difference was not significant (P = .1, fig. 5). Brain/plasma ratios were significantly different between the two groups (0.8 ± 0.6 vs. 3.8 ± 0.5, P < .001), reflecting primarily the increase in the plasma nicotine concentration in the CMUNic-KLH group. Protein binding of nicotine in plasma determined by equilibrium dialysis was significantly higher in the CMUNic-KLH group (83.4 ± 6.8% vs. 16.4 ± 14.2%, P < .001). Unbound nicotine concentrations did not differ (table 1).

View larger version (14K): [in this window] [in a new window]   Fig. 4.   Mean ± S.D. venous nicotine concentrations measured by gas chromatography after administration of nicotine 0.03 mg/kg i.v. over 10 sec. Plasma nicotine concentrations were significantly higher in rats immunized with CMUNic-KLH than in those immunized with KLH alone (P < .001).

View larger version (22K): [in this window] [in a new window]   Fig. 5.   Nicotine concentrations in brain 40 min after nicotine administration, and the brain/plasma ratio at the time. The difference in brain concentrations was not significant (P = .1). The marked increase in the brain/plasma ratio reflected primarily the increased plasma nicotine concentration. * P < .001.

	Discussion

Abstract Introduction Materials & Methods Results Discussion References

Previous studies have shown that immunization can modify the behavioral effects of heroin (Bonese et al., 1974; Killian et al., 1978) and cocaine (Carrera et al., 1995; Fox et al., 1996), presumably by binding drug in serum (and perhaps elsewhere) and preventing its distribution to brain. However, limited information is available on the quantitative requirements for immunization to alter drug distribution, such as the antibody affinity, binding capacity and extent of drug binding necessary to reduce drug concentrations in brain. The current study provides preliminary information relevant to these considerations for the use of immunization to alter nicotine distribution. These data should be useful in the design and screening of immunogens and immunization regimens intended to alter the behavioral effects of nicotine.

The rat was chosen for this study because of the ease of immunization (Carrera et al., 1995) and because nicotine pharmacokinetics in rats is quite similar to that in humans. Both are characterized by a high systemic clearance (human 91 vs. rat 152 l/min · kg), short elimination half-life (human 119 vs. rat 66 min), relatively large volume of distribution (human 2.6 vs. rat 4.1 l/kg), low protein binding in serum (human 4.9 vs. rat 9.5%) and limited renal excretion of unchanged drug (human 16 vs. rat 17%) (Benowitz et al., 1982; Benowitz et al., 1989; Plowchalk et al., 1992; Miller et al., 1977). The fraction of a dose converted to cotinine is lower in rats than in humans (27 vs. 70%), and the fraction converted to nicotine-N-oxide is probably greater (Jacob et al., 1981; Shulgin et al., 1987), but this should be of little consequence for this model so long as the antibodies elicited do not appreciably bind metabolites.

The nicotine dosing regimen used in this study was designed to simulate several critical features of nicotine distribution observed with cigarette smoking in humans. The nicotine dose was equivalent to that delivered by two cigarettes (Benowitz and Jacob, 1984), and it produced venous plasma nicotine concentrations in the 10 to 40-ng/ml range observed in regular smokers (Benowitz et al., 1983). The administration of nicotine by rapid i.v. injection was intended to simulate the rapid absorption of nicotine from the lungs associated with smoking. In humans, this results in arterial nicotine concentrations than are up to 10-fold higher than venous nicotine concentrations 1 min after a single cigarette (Henningfield et al., 1993) because of rapid delivery of the nicotine bolus from the pulmonary circulation to the heart. This results in the rapid delivery of these very high arterial concentrations of nicotine to the brain. This aterio-venous concentration difference, although not identical in time course, was also observed in our model. Because immunotherapy to reduce the behavioral effects of nicotine must overcome this rapid delivery of nicotine to the brain, it is a critical feature of a model for this intervention.

The immunogen used in this study was originally designed to produce antibodies for a radioimmunoassay to measure nicotine concentrations. Derivatization at the 6-position of the pyridine ring was chosen because it is remote from the major sites of nicotine metabolism (to cotinine and nicotine-N-oxide) on the N-methyl pyrrolidine ring, an approach that has also been used by others to produce antibodies specific to parent drug (Castro et al., 1980). This strategy appeared to be successful, in that cross-reactivity with these metabolites in the serum of immunized rats was negligible. However, the ELISA binding activity of serum was inhibited more readily by the complete immunogen CMUNic than by nicotine, which indicates that serum antibodies recognized the carboxymethylureido linker. These data suggest that derivatization of nicotine at the 6-position produces antibodies that recognize parent compound and not metabolites, but that a linker other than CMUNic would be preferable to improve antibody specificity further. The use of a chiral immunogen to target the naturally occurring and active ()-nicotine isomer, rather than the racemic immunogen used in this study, might also improve antibody specificity.

The affinity of serum antibodies for nicotine, measured using a soluble radioimmunoassay, was modest (2.4 ± 1.6 × 10⁷ M¹). The affinity of drug-specific antibodies used by Fox et al. (1996) to alter cocaine self-administration was reported to be higher, with a range of 4 × 10⁷ to 2 × 10⁹ M¹. Additional data on the required affinity are available from the literature on the use of passive immunization as a means of treating drug overdose. In this setting, heterologous drug-specific antibodies are administered rapidly (typically over minutes) to animals or patients with acute drug toxicity. The drug-specific antibodies bind drug in serum, reduce the unbound drug concentration in serum and provide a concentration gradient for drug to move out of target organs. If a sufficient dose of antibody is administered, toxicity can be reduced or completely reversed. Digoxin (Smith et al., 1982) and colchicine (Baud et al., 1995) toxicity have been treated in patients in this manner, and phencyclidine (Valentine et al., 1996), desipramine (Brunn et al., 1992) paraquat (Chen et al., 1994) and ricin (Houston, 1982) toxicity have been reversed in animals. Most antibodies used to treat drug overdose have had affinities for drug of 10⁸ M¹ or higher, (10¹⁰ M¹ for digoxin, 2 × 10¹⁰ M¹ for colchicine, 5.6 × 10⁸ M¹ for phencyclidine, 3 × 10⁸ M¹ for desipramine), although a direct comparison of antibodies with differing affinities is not available to address the question of what affinity is actually required (Smith et al., 1982; Baud et al., 1995; Valentine and Owens, 1996).

Despite the modest affinity of the elicited nicotine-specific antibodies, immunization in the current study had a substantial effect to nicotine distribution. Nicotine binding in plasma was increased from 16.4% in control rats to 83.4% in the CMUNic group. As a result, total nicotine concentrations in plasma were 4 to 6-fold higher in the CMUNic group. Using the plasma nicotine concentration difference between the CMUNic and control groups, and the estimated plasma volume of the rat, a calculated 9% of the administered nicotine dose was bound by drug-specific antibodies in serum. Although serum represents the largest reservoir for IgG (Harlow and Lane, 1988), this figure could be higher if some binding takes place outside of serum. The nicotine binding capacity of plasma calculated from radioimmunoassay data (210 ng/ml) was 3-fold higher than that which was measured in vivo (67 ng/ml). This probably reflects the modest affinity of the serum antibodies for nicotine resulting in incomplete occupancy of antibody by drug.

The nicotine-specific IgG concentration achieved in immunized rats was 0.1 mg/ml, or about 1% of total IgG. This is comparable to the 0.08 mg/ml produced by passive immunization with cocaine-specific antibodies and found to be effective in abolishing cocaine self-administration (Fox et al., 1996). Although they are typically lower, antigen-specific IgG concentrations in this range have been reported in humans (Weiss et al., 1995; Janoff et al., 1991), which supports the clinical relevance of this rat model.

The brain/plasma nicotine ratio in control animals of 3.8 was similar to the value of 3.3 reported previously in rats 60 min after nicotine dosing (Chowdhury et al., 1993). The brain/plasma nicotine ratio was significantly lower in the CMUNic group (0.8), but this was primarily due to the increase in the plasma nicotine concentration. Brain nicotine concentrations in the two groups were not significantly different. The brain nicotine concentration in the CMUNic group was, however, 13.8% lower than in controls, a result similar to the estimate of the amount of nicotine bound in plasma. It is possible that the brain nicotine concentration was in fact somewhat lower in the CMUNic animals but that the limited power of this small study was insufficient to demonstrate this effect. It is not clear whether such a small reduction in brain nicotine concentration would be sufficient to produce behavioral effects. Although this possibility should be studied, better immunogens that produce higher-affinity antibodies would clearly be desirable. The more pronounced (approximately 50%) reductions in brain cocaine concentrations reported by others to result from immunization (Carrera et al., 1995; Fox et al., 1996) may reflect higher affinities or serum titers of these drug-specific antibodies, differences in drug doses or differences in sampling times. In particular, measuring the brain nicotine concentration earlier than 40 min would be of interest.

In summary, the effects of active immunization on nicotine distribution were studied in a rat model that simulated the arterio-venous difference in nicotine levels produced by cigarette smoking and resulted in clinically relevant plasma nicotine concentrations. Immunization markedly increased the binding of nicotine in plasma, an estimated 9% of the administered dose being bound by antibody. These effects that were observed despite the use of an immunogen that produced antibodies of only modest affinity suggests that immunotherapy, particularly with improved immunogens, has the potential to alter substantially the distribution of nicotine and perhaps alter its behavioral effects.