Introduction

Top Abstract Introduction Materials & Methods Results Discussion References

Many chemicals and trace elements are sequestered in human hair after exposure to cocaine (Cone et al., 1991; Nakahara and Kikura, 1994), opioids (Cone, 1990; Kintz and Mangin, 1993), amphetamine (Kintz et al., 1995; Nakahara, 1995) and nickel, lead and strontium (Sky-Peck, 1990; Yoshinaga et al., 1993; Ping and Xingquan, 1993). The exact binding sites in hair and the physicochemical properties of these sites have not been elucidated because of the complex structure of hair. Hair is an epidermal tissue derived from follicles located in the dermis of skin (Montagna and Van Scott, 1958; Harkey, 1993). Blood delivers oxygen and nutrients to the hair follicle, which contains metabolically active cells (Hashimoto et al., 1994). Dermal papillae cells in the follicle differentiate into three basic layers of hair. The cuticle is the outermost layer of hair, the cortex is located beneath the cuticle, and the medulla is the innermost layer (Montagna and Van Scott, 1958; Swift, 1981; Kassenbeck, 1981). The cortex, medulla and cuticle are composed mainly of fibrous and matrix proteins (65%-95%), lipids (1%-9%) and water (Harkey, 1993; Dekio and Jidoi, 1988; 1990). Fibrous proteins have an alpha helix structure and are embedded in nonhelical, matrix proteins (Baden, 1981; 1989; Gillespie and Marshall, 1981). These proteins are collectively referred to as keratin (Matoltsy, 1967). Melanins are natural pigments present in hair and skin (Fitzpatrick et al., 1958; Thody et al., 1991; Barnicot and Birbeck, 1958; Montagna et al., 1994). These pigments can be synthesized from a variety of substrates, including dopa, dopamine, tyrosine, and catechol (Yu and Van Scott, 1973; Swan, 1974). Melanin is considered to be made up of indole quinone polymers that are synthesized in melanocytes derived from the neural crest (Mason, 1959; Fitzpatrick et al., 1958; Ortonne and Prota, 1993). After synthesis, vesicles containing melanin are transferred from melanocytes to cortical and medullary cells of hair. The concentration and type of melanin in hair determine the color (Ortonne and Prota, 1993; Cesarini, 1990; Birbeck and Barnicot, 1959; Barnicot and Birbeck, 1958). Dark hair generally contains a higher concentration of melanin than light hair. Lipids are also present in hair and include free fatty acids, cholesterol sulfates and ecosinoic fatty acids (Wertz and Downing, 1988; 1989).

The disposition and retention of chemicals in hair may occur by multiple processes and may depend on certain morphological and structural characteristics of hair. When trace metals and drugs are ingested, these chemicals enter the general circulation and may be incorporated into the structure of hair during the synthesis of cellular components (Henderson, 1993; Nakahara et al., 1992; Cone, 1995). Chemicals may also enter hair from sweat and sebum that bathe the hair shaft (Henderson, 1993; Faergemann et al., 1990; Blank and Kidwell, 1993). Furthermore, airborne chemicals, including cocaine vapor and trace metals in the environment, may be deposited on and retained by hair (Blank and Kidwell, 1993; Kopito and Shwachman, 1974; Kidwell and Blank, 1992; Wang and Cone, 1995). Hair proteins, melanins and lipids contain many potentially reactive groups capable of binding chemicals that enter hair cells (Montagna et al., 1994; Ward and Lundgren, 1954). Drugs and metals may bind by forming weak electrostatic interactions, hydrophobic attractions and ionic bonds with hair components (Larsson and Tjalve, 1978; 1979; Montagna et al., 1994). However, hair types in the general population differ in melanin content (Thody et al., 1991; Cone and Joseph, 1996), in protein content (Gerhard, 1987; Gerhard and Hermes, 1987; Dekio and Jidoi, 1990), in structure (Hrdy, 1973), in morphology (Steggerda and Seibert, 1941; Lindelof et al., 1988; Noback, 1951) and in chemical treatments with cosmetic products. These differences may affect the binding of drugs to hair. The incorporation of many drugs into hair including cocaine (Joseph et al., 1996), methadone (Green and Wilson, 1996) and codeine and morphine (Gygi et al., 1996) has been reported to differ with hair color. For these drugs, incorporation into dark hair is generally much greater than into light hair. Treatment of hair with cosmetic products such as peroxide bleaches and coloring agents also has been reported to decrease cocaine binding (Joseph et al., 1996) and to affect trace element concentrations (Sky-Peck, 1990) in hair. The effect of differences between hair types on the extent of drug incorporation into hair is an important issue, because hair is currently being collected and analyzed to identify drug use by individuals in the workplace (Karch, 1996; Cook et al., 1995) and in many other applications (Feucht et al., 1994; Forman et al., 1994; Huestis, 1996; Callahan et al., 1992). A major concern is that hair test results may not be consistent or impartial for individuals with different hair types. Therefore, an understanding of the nature of drug binding to hair is needed in interpreting hair test results.

In vitro binding techniques have been used often to evaluate the mechanism(s) by which drugs produce pharmacological effects (Deecher et al., 1995; Katsumata et al., 1995; Hustveit et al., 1995). The binding of drugs of abuse to hair has no known pharmacological effects; however, specific binding sites in hair have been identified in vitro for cocaine (Joseph et al., 1996) and for codeine and morphine (Gygi et al., 1996). Support for the use of in vitro studies to evaluate drug binding to hair is based on the observation that in vitro radiolabeled codeine and morphine binding to rat hair is similar to in vivo drug incorporation (Gygi et al., 1996). Similarities in the extent of in vivo and in vitro drug incorporation into dark and light hair also are evident, although in vivo data are limited (Reid et al., 1994; Kidwell and Blank, 1992; Reuschel and Smith, 1991; Joseph et al., 1996). The present study characterized binding sites in human hair by using in vitro techniques to determine whether selective, high-affinity cocaine binding to hair occurs that is similar to drug binding to receptors in biological tissues. Cocaine binding also was evaluated with light and dark male and female ethnic hair types to determine how hair type affects cocaine binding.

	Materials and Methods

Top Abstract Introduction Materials & Methods Results Discussion References

Hair specimens. Hair was collected from staff at the Addiction Research Center and from barber shops in the Baltimore and Washington area. Specimens were stored at 30°C in plastic bags until the time of analysis. Hair was initially analyzed by gas chromatography and mass spectrometry for cocaine and metabolite content according to a previously published procedure (Cone et al., 1991). Only drug-free specimens were utilized in binding assays. The color of hair specimens was classified visually as light (blond) or dark (black and brown). Only hair specimens that could be clearly distinguished as black and brown or as blond were included. Dark and light hair specimens used in saturation and competition assays had not been subjected to previous chemical treatments other than Head and Shoulders shampoo. A history of hair treatment was not available for specimens included in kinetic assays; in assays to determine the effects of temperature, NaCl, and pH on [³H]cocaine binding; or in a population study to determine [³H]cocaine binding to 156 individual hair specimens.

Preparation of hair suspensions. Hair suspensions were prepared according to a previously published procedure (Joseph et al., 1996). Briefly, hair specimens were placed in filtration columns and washed with tap water (3 × 10 ml). Specimens were dried to a constant weight, and 15 to 25 mg of hair was weighed in 5-ml Elkay polypropylene tubes. Hair was cut into segments less than or equal to 0.5 cm with surgical scissors, and Kimble 2.5-mm borosilicate beads were added to the tubes until one-third of the volume of the tube was filled. The beads were added to homogenize specimens by impact during operation of the reciprocating BioSpec Mini-Beadbeater-8 cell disrupter. Tubes were capped and placed in a BioSpec Mini-Beadbeater-8 cell disrupter (Bartlesville, OK) modified to accommodate 5-ml tubes. The dial was set at 50% of maximum speed (1400 oscillations per minute), and specimens were homogenized for 3 min. Tubes were removed, shaken manually and homogenized again for 3 min. The homogenization process generally pulverized hair fragments into fine particles less than 1 mm long. Many hair particles were observed to be less than 20 µm in length by microscopic examination and measurement. However, some hair particles were approximately 1 mm in length. These particles usually represented less than 10% of the amount of hair initially prepared for homogenization. After homogenization, 3.0 ml of 50 mM Tris-HCl buffer (pH 7.4) was added to tubes that were shaken in the homogenizer for 5 s. The suspension was transferred to 50-ml Elkay tubes, and this procedure was repeated two additional times. Hair suspensions were diluted with 50 mM Tris-HCl buffer (pH 7.4) to the desired concentration.

Binding assays. Dark and light hair homogenates in 50 mM Tris-HCl buffer (pH 7.4) were prepared in duplicate or triplicate in all assays to measure [³H]-()-cocaine (levo-[benzoyl-3,4-³H(N)]) binding to hair. Nonspecific binding was determined by preparing hair suspensions with [³H]cocaine and 10 µM ()-cocaine in all assays except the population study, in which nonspecific binding was defined by 100 µM ()-cocaine. Saturation assays were performed with light hair obtained from a male and female Caucasoid and dark hair collected from a female Africoid, a male Africoid and a male Mongoloid. In saturation assays, 1-ml hair suspensions were prepared with 0.4 to 1.5 mg of hair and 1 nM to 3 µM [³H]cocaine (1.0 Ci/mmol). Suspensions were incubated in a shaking water bath at 25°C for 1 h before filtration.

Statistics. Data from association and dissociation assays were analyzed with a weighted, iterative, nonlinear curve-fitting program (Biosoft® Kinetic program, London, UK). Data from saturation experiments were analyzed with Biosoft EBDA and Ligand curve-fitting programs as described by Munson and Rodbard (1980). An F test of variance was used to determine whether data in binding assays fit a monoexponential or a biexponential kinetic model. The effect of hair color on [³H]cocaine binding was evaluated by multivariate analysis that included 123 dark and 33 light hair specimens. Logarithmic transformation of binding data was performed to compensate for unequal variances between groups of light and dark hair specimens. The number of light hair specimens was weighted by a factor of 3.7 to compensate for unequal group sizes. Multivariate and univariate analyses were performed to determine the effects of ethnicity and sex on [³H]cocaine binding to hair. Data obtained for radioligand binding to light female Caucasoid hair specimens were excluded from the analyses to determine the effects of ethnicity and sex on cocaine binding to hair, because representative groups of light male Caucasoid and light male and female Africoid hair specimens were unavailable for analysis. The effect of individuals' ages on [³H]cocaine binding was evaluated by analysis of covariance. Multivariate analysis was performed to determine whether light and dark hair specimens differed in response to chemical treatments and temperature changes. Paired t tests were used to determine whether chemical treatments and temperature changes affected mean [³H]cocaine binding for treatment groups compared with a control group comprising hair suspensions prepared at pH 7.4 and incubated at 25°C.

Drugs. The following drugs and chemicals were used in this study: [³H]-()-cocaine, specific activity 30.0 Ci/mmol, (New England Nuclear, Boston, MA), ()-cocaine HCl and mazindol (Sigma Chemical Company, St. Louis, MO), Win 35,428, RTI-51 and (+)-cocaine HCl (Research Triangle Institute, Research Triangle Park, NC) and Poly-fluor liquid scintillation cocktail (Packard Chemical Co., Meriden, CT). All other chemicals were reagent grade.

	Results

Top Abstract Introduction Materials & Methods Results Discussion References

Reversible [³H]cocaine binding in hair. Binding of [³H]cocaine to dark and light hair was evaluated by preparing suspensions that contained from 5 nM to 3 µM radioligand and from 0.1 to 1.5 mg of hair/ml. Specimens were incubated at 25°C for 1 h, and filtration was used to separate bound and unbound [³H]cocaine to determine total, nonspecific and specific binding to hair. Figure 1 illustrates [³H]cocaine binding to light and dark hair. Specific binding was initially greater than nonspecific for dark hair, but specific binding began to plateau when hair suspensions contained 2 µM [³H]cocaine. Specific binding to light hair was less than nonspecific binding at all [³H]cocaine concentrations. When hair suspensions contained a fixed radioligand concentration (500 nM), specific binding increased linearly with increasing concentration of dark and light hair (fig. 2).

View larger version (14K): [in this window] [in a new window]   Fig. 1.   [3H]cocaine binding to hair. Dark hair was obtained from a male Mongoloid and light hair from a male Caucasoid. Total binding was determined by preparing hair suspensions that contained from 5 nM to 3 µM [3H]cocaine (1 Ci/mmol). Nonspecific suspensions also contained 10 µM ()-cocaine. Suspensions were incubated at 25°C for 1 h before filtration. Each point for total and nonspecific binding represents the mean for two determinations. Specific binding is the difference between mean total and mean nonspecific binding.

View larger version (15K): [in this window] [in a new window]   Fig. 2.   [3H]cocaine specific binding versus hair concentration. Dark hair was obtained from a male Mongoloid and light hair from a male Caucasoid. Hair suspensions were prepared with 5 nM [3H]cocaine (30 Ci/mmol) and also with 10 µM ()-cocaine for nonspecific binding determination. The concentration of hair in suspensions ranged from 0.1 to 1.5 mg/ml. Specimens were incubated at 25°C for 1 h before filtration. Suspensions that contained hair concentrations greater than 1.5 mg/ml could not be filtered. Specific binding represents the difference between mean total and mean nonspecific binding.

[³H]cocaine association kinetics in hair. Association time course assays were performed by preparing dark and light hair suspensions with 2000 nM [³H]cocaine and incubating specimens at 25°C for intervals ranging from 1 to 120 min before filtration. Specific binding data were analyzed with the Biosoft Kinetic program to obtain binding parameters. Data analysis indicated that a biexponential association model was statistically preferred (P < .05) to a monoexponential association of radioligand to binding sites (fig. 3). The k_obs1 and k_obs2 values for dark and light hair are listed in table 1. Binding was characterized by rapid association at site 1, followed by a slower phase at site 2, in both hair types. The k_obs values also were similar for dark and light hair. Overall binding equilibrium for both hair types was attained within 60 min. At equilibrium, 26% ± 1% (mean ± S.E.M., n = 4) of [³H]cocaine was bound at site 1 and 74% ± 1% at site 2 in dark and light hair specimens.

View larger version (14K): [in this window] [in a new window]   Fig. 3.   Association time course for [3H]cocaine specific binding in hair. Dark hair was pooled from four male Africoids, three female Africoids and three female Caucasoids. Light hair was pooled from six female Caucasoids. Suspensions were prepared in duplicate with 2000 nM [3H]cocaine (1 Ci/mmol) and incubated at 25°C. Nonspecific binding suspensions also contained 10 µM ()-cocaine. Suspensions were filtered at intervals ranging from 1 to 120 min after the addition of [3H]cocaine.

[³H]cocaine dissociation kinetics in hair. Dissociation of [³H]cocaine from dark and light hair was determined by incubating radioligand suspensions for 1 h at 25°C followed by 10-fold dilution of specimens in buffer that contained 100 µM ()-cocaine. Filtration was performed at intervals ranging from 1 to 80 min after dilution. Specific binding data were analyzed by the Biosoft Kinetic program to obtain k_1i values. Analysis of the data indicated that a biexponential dissociation model was statistically preferred (P < .05) to a monoexponential model. Figure 4 illustrates dissociation of [³H]cocaine from hair. Dissociation of [³H]cocaine from dark and light hair was similar at site 1, although k₁₂ was more rapid in dark hair than in light hair (table 1). Nonspecific binding also decreased for dark and light hair specimens after dilution. For specimens filtered 80 min after dilution with 100 µM ()-cocaine, nonspecific binding in dark and light hair decreased by 81% ± 1% (mean ± S.E.M., n = 4) compared with specimens filtered at time 0. 

View larger version (13K): [in this window] [in a new window]   Fig. 4.   Dissociation of [3H]cocaine from hair. Dark hair was pooled from four male Africoids, three female Africoids and three female Caucasoids. Light hair was pooled from six female Caucasoids. Light hair suspensions were prepared with 50 nM [3H]cocaine (30 Ci/mmol), and dark hair suspensions contained 40 nM [3H]cocaine (30 Ci/mmol). Nonspecific binding was defined by 10 µM ()-cocaine. Suspensions were incubated for 1 h at 25°C, and then dissociation of specifically bound [3H]cocaine was determined by 10-fold dilution of suspensions in 50 mM Tris-HCl buffer (pH 7.4) that contained 100 µM ()-cocaine. Filtration was performed at intervals ranging from 1 to 80 min after dilution. Specific binding represents the difference between mean total and mean nonspecific binding for two determinations.

1₂

1₂

1₁

Scatchard analysis of [³H]cocaine binding sites. Dark and light hair suspensions were prepared with 5 nM to 3 µM [³H]cocaine. Specimens were filtered after incubation for 1 h, and specific binding was determined. These data were used to construct Scatchard plots to determine the affinity and density of binding sites in hair. Scatchard plots were curvilinear (fig. 5), and regression analysis demonstrated the presence of two affinities in hair for [³H]cocaine. The K_d values are listed in table 2. The K_d1 value was lower in one light hair specimen than in dark hair specimens. However, specific binding was not detected for another light hair specimen. Scatchard analysis also indicated that the density of low-affinity sites was greater than that of high-affinity sites in dark and light hair. However, the density of high- and low-affinity sites in dark hair ranged from 5- to 43-fold greater than in light hair.

View larger version (15K): [in this window] [in a new window]   Fig. 5.   Scatchard analysis of specific [3H]cocaine binding in hair. Specimens were prepared as described in figure 1. Specific binding data were analyzed by the Biosoft EBDA and Ligand program using iterative curve-fitting methods. A two-site binding model was statistically preferred (P < .01) compared with a single-site model for light and dark hair.

Effect of NaCl, temperature and pH on [³H]cocaine binding. Binding of [³H]cocaine under different experimental conditions was determined for dark and light hair suspensions that contained 500 nM ³H()cocaine. Control suspensions for each hair specimen were prepared with radioligand in 50 mM Tris-HCl buffer at pH 7.4 and incubated at 25°C for 1 h before filtration. Figure 6 illustrates the mean responses for light and dark hair relative to controls. In figure 6A, specific and nonspecific binding were not affected by changes in incubation temperature.

View larger version (16K): [in this window] [in a new window]   Fig. 6.   The effects of temperature, pH and NaCl on [3H]cocaine binding in hair. Dark hair specimens were collected from two male Africoids, one male Caucasoid, two female Africoids and one female Caucasoid. Light hair was collected from six female Caucasoids. Each hair specimen was divided and placed in a control group and in each treatment group. Control suspensions were prepared at pH 7.4 in 50 mM Tris-HCl buffer and incubated with [3H]cocaine for 60 min at 25°C before filtration. Multivariate analysis demonstrated that the effects of pH, temperature and NaCl on specific and nonspecific binding were similar for light and dark hair specimens (F = 0.49, P > .1). Each column represents the mean percent response for all hair types relative to their individual controls. a) Specimens were heated to 80°C for 30 min, prepared with 500 nM [3H]cocaine (1 Ci/mmol), and incubated for 1 h at 25°C before filtration. Separate suspensions were prepared with 500 nM [3H]cocaine (1 Ci/mmol) and incubated for 1 h at 4°C before filtration. b) Suspensions were prepared at pH 10.5 and pH 3.0. [3H]cocaine (500 nM, 1 Ci/mmol) was added and suspensions were incubated for 1 h at 25°C before filtration. c) Suspensions were prepared with 500 nM [3H]cocaine (1.0 Ci/mmol) and 0.49 M NaCl. Suspensions were incubated for 1 h at 25°C before filtration. Bars represent S.E.M. ** A significant (P < .05) difference between control and treatment groups was determined with paired t tests.

Competition for binding sites. We further characterized binding sites for [³H]cocaine in competition studies by determining the potency of (+)-cocaine, RTI-51, WIN 35,428 and mazindol at inhibiting radioligand specific binding to hair. Dark and light hair suspensions were prepared with 5 nM [³H]cocaine and from 1 nM to 100 µM competing ligands. Suspensions were incubated for 1 h at 25°C, and specific binding was determined after filtration. The results are listed in table 3. ()-Cocaine, (+)-cocaine and RTI-51 competed with [³H]cocaine at high- and low-affinity binding sites in hair. At both sites, the potency of ()-cocaine was approximately 10-fold greater than that of (+)-cocaine (fig. 7). RTI-51 was less potent than ()-cocaine. Win 35,428 and mazindol competed with [³H]cocaine at only one group of binding sites, and these ligands had a potency similar to or less than that observed for ()-cocaine at low-affinity binding sites in hair.

View larger version (18K): [in this window] [in a new window]   Fig. 7.   Stereoselective [3H]cocaine binding in hair. Dark hair was obtained from a male Mongoloid and light hair from a male Caucasoid. Hair suspensions were prepared with 5 nM [3H]cocaine (30 Ci/mmol) and from 1 nM to 100 µM (+)-cocaine or ()-cocaine. Suspensions were incubated for 1 h at 25°C before filtration. Specific binding represents the difference between mean total and mean nonspecific binding for two determinations.

Population study. Dark and light hair specimens were collected from 156 male and female Africoid and Caucasoid volunteers. [³H]Cocaine binding was determined with hair homogenates in suspensions that contained 500 nM [³H]cocaine. Nonspecific binding was determined by adding 100 µM ()-cocaine to suspensions. The results are shown in figure 8. Logarithmic transformation of binding data was performed to compensate for unequal variances between groups of hair specimens. Multivariate analysis indicated a significant effect (F = 77.3, P < .001) for hair color on specific and nonspecific radioligand binding to 156 hair specimens. Mean specific binding was 13,131 ± 1463 dpm/mg of hair (mean ± S.E.M., n = 123) for dark hair and 1394 ± 275 dpm/mg of hair (mean ± S.E.M., n = 33) for light hair specimens. Mean nonspecific binding was 4596 ± 371 dpm/mg of hair (mean ± S.E.M., n = 123) for dark hair and 1661 ± 203 dpm/mg of hair (mean ± S.E.M., n = 33) for light hair specimens.

View larger version (16K): [in this window] [in a new window]   Fig. 8.   Comparison of specific and nonspecific [3H]cocaine binding to 156 dark and light hair specimens as described in table 4. Multivariate analysis indicated that specific and nonspecific [3H]cocaine binding were significantly greater (F = 77.3, P < .001) for dark hair compared with light hair. Bars represent mean binding for dark and light groups of hair specimens.

	Discussion

Top Abstract Introduction Materials & Methods Results Discussion References

This study demonstrated selective and reversible cocaine binding to hair. Specific binding generally defines the reversible association of a ligand to a binding site. There are many examples of specific binding of drugs to a variety of biological specimens, including brain (Calligaro and Eldefrawi, 1988), lymphocytes (Le Fur et al., 1980), placental tissue (Aaltonen et al., 1983), albumin (Adal et al., 1995), skin (Choy et al., 1995) and spermatozoa (Aanesen et al., 1995; Yazigi et al., 1991). In studies of putative receptors, the measurement and comparison of specific binding under different experimental conditions (e.g., variations in tissue concentration, competitor and temperature) have provided insight into the mechanisms of ligand binding. This approach was employed in the present study to characterize cocaine binding sites in hair and to compare these sites with those present in biological tissues.

In typical receptor-ligand binding studies, an increase in the density of high-affinity binding sites results in an increase in ligand binding. In the present study, a linear increase in cocaine specific binding occurred with increasing concentration of dark and light hair (fig. 2). Similarly, Madras et al. (1989) and Reith et al. (1980) demonstrated a linear increase in specific cocaine binding with an increase in brain homogenate concentration. Braestrup and Squires (1977) characterized specific benzodiazepine binding sites in rat brain and reported a linear increase in specific binding of diazepam with increasing tissue concentration. Le Fur et al. (1980) examined specific binding of the potent dopaminergic antagonist spiroperidol to rat lymphocytes and found a linear relationship between drug binding and cell concentration. Ligand binding to tissues is determined by the density of sites and by the affinity state of binding sites. In the present study, the affinity of binding sites in hair for cocaine was initially examined in kinetic studies.

Evaluation of the association time course for cocaine binding to specific sites in hair demonstrated that binding approached equilibrium at approximately 1 h and was best represented by a biexponential association model. The k_obs values were similar for dark and light hair and ranged from 0.97 to 1.1 min¹ for k_obs1 and 0.035 min¹ for k_obs2. Calligaro and Eldefrawi (1988; 1987) reported a k_obs for cocaine binding to a single site in whole-brain synaptic membranes (1667 min¹) and in striatal membranes (4.54 min¹). Steady state was reached within 60 s for cocaine binding to synaptic and striatal membranes. Reith et al. (1981) described a rapid association of cocaine to sites in rat brain membranes (k_obs = 10 min¹) with attainment of equilibrium at approximately 10 s. Therefore, the overall time course for cocaine binding to hair in the present study was generally longer than that for cocaine binding to sites in other tissues.

Dissociation of cocaine from hair was best represented by a biexponential model. The k_1i values obtained for light and dark hair ranged from 0.009 to 0.01 min¹ for k₁₁ and from 0.49 to 0.16 min¹ for k₁₂. In comparison, the k₁₁ values in the present study were lower than the single k₁₁ value (0.026 min¹) determined for cocaine dissociation from liver membranes (Calligaro and Eldefrawi, 1987). Calligaro and Eldefrawi (1988) also reported the presence of a rapid k₁₁ (16.6 min¹) and a slower k₁₂ (0.86 min¹) for cocaine dissociation from striatal membranes. Reith et al. (1980) reported dissociation of saturably bound [³H]cocaine from mouse brain membranes at 1 min after sample dilution. Calligaro and Eldefrawi (1987) reported that cocaine dissociation from rat brain membranes was too rapid to measure. The slow dissociation of cocaine from hair in the present study was considerably less rapid compared with reports of cocaine dissociation from other tissues.

The presence of multiple affinities in hair for cocaine was further investigated. Scatchard plots were curvilinear and demonstrated the presence of high and low affinities for cocaine binding to hair. The high-affinity cocaine binding reported in the present study for two hair specimens is comparable to that described by Calligaro and Eldefrawi (1987) for whole brain (K_d = 16 nM), striatum (K_d = 17 nM), occipital cortex (K_d = 17 nM) and frontal cortex (K_d = 14 nM). For these tissues, a second affinity also was detected, and K_d values ranged from 68 to 660 nM. Other K_d values reported for cocaine binding to brain components range from 29 to 900 nM (Reith et al., 1980; 1981; 1986; Calligaro and Eldefrawi, 1988). These values are similar to or less than the K_d values determined in the present study for low-affinity sites in hair.

Association and dissociation rate constants were used to determine kinetic K_d values for comparison with those obtained by Scatchard analyses of binding data. The kinetic K_d estimates were slightly lower than those obtained by Scatchard analyses. Different hair specimens were used in binding assays for kinetic and Scatchard analyses, which may explain the differences in K_d values. However, the rapid dissociation rate constant determined in kinetic analyses may have been due to reversible nonspecific binding that could have affected kinetic K_d estimates. More than 80% of nonspecific binding in dissociation assays was reversible when suspensions were diluted with 100 µM ()-cocaine and filtered after 80 min. In addition, the rate of diffusion of [³H]cocaine and ()-cocaine into hair and the exposure of binding sites to these ligands may have affected binding in hair suspensions filtered before the attainment of binding equilibrium in association and dissociation assays.

The diffusion of cocaine and other drugs within hair has received limited attention. Potsch and Moeller (1996) examined the diffusion of rhodamine B dye into hair fibers. They reported that penetration of the dye occurred predominantly through the nonkeratinous regions of the cell membrane complex. These investigators also indicated that complete penetration of the dye into the inner hair structure, including the cortex and medulla, occurred at 30 to 60 min after dye exposure, depending on temperature. Similarly, Skopp et al. (1996) reported that codeine, dihydrocodeine and morphine diffuse through the nonkeratinous regions in whole hair fibers. The time required for diffusion through 8.5-cm segments of hair ranged from 24 to 96 h. Some hair fragments in the present study were larger than 1 mm in length. In kinetic assays, diffusion-limited processes may have affected binding measures when hair suspensions were filtered within minutes after the addition of [³H]cocaine in association assays and of ()-cocaine in dissociation assays. In contrast, a steady-state binding equilibrium was attained before the filtration of suspensions in assays used to obtain data for Scatchard analyses.

The nature of specific binding was further investigated in competition studies. The natural isomer ()-cocaine was approximately 10-fold more effective than (+)-cocaine at inhibiting [³H]cocaine binding. There are many examples of stereoselective drug binding to biological tissue components (Blank et al., 1986; Valenzuela et al., 1995; del Pozo et al., 1996). Ritz et al. (1987) examined the potencies of cocaine and other drugs at inhibiting dopamine uptake in rat striatal tissue. ()-Cocaine was 213-fold more potent than (+)-cocaine at inhibiting dopamine uptake. The potency of ()-cocaine also was 116 fold-greater than that of (+)-pseudococaine at inhibiting dopamine uptake. Stereoselective cocaine binding has been described in cortical and striatal tissue, in which ()-cocaine was approximately 10-fold more potent than (+)-pseudococaine at inhibiting binding of radiolabeled dopamine and serotonin (Reith et al., 1986). Calligaro and Eldefrawi (1988) reported that ()-cocaine was 100-fold more effective than (+)-pseudococaine at inhibiting cocaine binding to receptors in striatal membranes and at blocking radiolabeled dopamine binding in striatal synaptosomes. The study of chiral drugs has been important in identifying the structure and subtypes of receptors in tissues (Barbier et al., 1995; Ritz et al., 1987; 1990a; 1990b). Small differences in amino acid composition can determine the specificity of a binding site. For example, muscarinic receptor subtypes can be differentiated on the basis of their stereoselectivity for chiral antagonists, including biperiden (Eltze and Figala, 1988) and rociverine (Barbier et al., 1995). Stereoselective [³H]cocaine binding to hair in the present study indicated that a specific configuration of functional groups forms the binding site. However, the magnitude of stereoselective [³H]cocaine binding was considerably less in hair than in brain tissue. This provides additional evidence that hair binding sites differ from those described in other tissues.

Cocaine binding to hair was reversible, stereoselective and saturable. These are characteristics of receptors that are present in biological tissues. However, binding sites in hair appear to be structurally distinct. Competition studies demonstrated that the potencies of RTI-51, mazindol and Win 35,428 at inhibiting [³H]cocaine binding were similar to or less than that of ()-cocaine. In contrast, RTI-51 (Stathis et al., 1995), WIN 35,428 (Reith et al., 1986; Scheffel et al., 1989) and mazindol (Ritz et al., 1987; 1990b) are much more potent than cocaine at binding sites in brain, which suggests that binding sites in hair differ from those in the CNS that are thought to mediate the behavioral effects of cocaine.

Joseph et al. (1996) suggested that melanin is the primary specific binding site in hair for cocaine. Melanin is a pigment that is reported to bind drugs in many tissues, including locus ceruleus and substantia nigra (Lindquist, 1972), eye (Potts, 1962; Atlasik et al., 1980), hair (Cone and Joseph, 1996) and inner ear (Conlee et al., 1989). The basic structure of melanin consists of indole quinone units linked to form polymers that are supported upon a protein matrix (Mason, 1959; Swan, 1974). Melanin can be denatured by free radicals and oxidizing agents but is stable at temperature extremes. Proteins, on the other hand, are often subject to heat and cold denaturation (Iwakura and Honda, 1996; Liu and Sturtevant, 1996; Lu et al., 1992; Privalov, 1992). Calligaro and Eldefrawi (1987) reported nearly complete elimination of specific cocaine binding as a result of heat denaturation and proteolytic digestion of liver and brain, which suggests that proteins comprise cocaine binding sites in these tissues. Reith et al. (1980) also reported that cocaine binding sites in cerebral cortex were affected by heat treatment, which resulted in decreased cocaine binding. Pfeil and Privalov (1976) described the effects of temperature and pH on the stability of the globular protein lysozyme. They reported that changes in enthalpy and entropy were related primarily to the effects of temperature on lysozyme structure and that the denaturational enthalpy change for lysozyme was a linear function of temperature. In the present study, cocaine binding to most hair specimens was not affected by extremes of heat and cold, which indicated that hair components other than proteins, such as melanin, are probable cocaine binding sites.

Further characterization of cocaine binding sites in hair was accomplished by determining the effects of chemical treatments on cocaine binding. Hair suspensions prepared at pH 3.0 and separate suspensions prepared at pH 10.5 bound significantly (P < .05) less cocaine than suspensions prepared at pH 7.4. Kidwell and Blank (1995) reported that drugs bind to hair by hydrophobic interactions when suspension pH is less than 4.0. The investigators indicated that at higher pH values, drugs bind to hair by ionic interactions. Proteins and melanin contain many functional groups, including carboxyl and phenolic groups, that may be affected by pH changes. Obika (1976) reported that binding of riboflavin to melanin decreased when specimens were prepared at pH values greater than or less than 7.0. Sarna et al. (1980) investigated the mechanisms of copper binding to natural and synthetic melanin by using electron paramagnetic resonance spectroscopy to characterize copper and melanin complexes. They reported that copper binds to monodentate carboxyl complexes and bidentate nitrogen-carboxyl complexes in melanin and synthetic melanin when placed in suspensions at pH less than 7.0. Binding occurred between copper and phenolic hydroxyl groups when suspensions were prepared at pH values greater than 7.0. Kozik et al. (1990) investigated the effect of pH on riboflavin binding to synthetic melanin. They determined that riboflavin binding to melanin increased when suspension pH increased from 5 to 9. They also reported that riboflavin binding to synthetic melanin was a reversible and saturable process.

In the present study, the functional groups in melanin of dark and light hair may bind cocaine through electrostatic forces. Changes in the ionization state of the functional groups with changes in pH may be responsible for decreased cocaine binding in suspensions prepared at pH 3.0 and pH 10.5. In suspensions at pH 3.0, more than 99% of cocaine is cationic, but functional groups in hair are primarily neutral (Kidwell and Blank, 1995). Binding in suspensions with pH below the isoelectric point of hair (4.0) is reported to occur mainly by weak hydrophobic interactions (Kidwell and Blank, 1995). At pH 10.5, functional groups in hair are mainly negatively charged, but only 1% of cocaine is cationic and capable of ionic interactions with functional groups in hair. In suspensions at pH 7.4, hair is negatively charged, and approximately 94% of cocaine is cationic. Therefore, binding by ionic mechanisms is expected to be optimal in suspensions at pH 7.4 compared with pH 3.0 and pH 10.5.

Cocaine binding to hair also was decreased by the addition of NaCl to hair suspensions at pH 7.4. Calligaro and Eldefrawi (1987) reported a similar decrease in cocaine binding to brain and liver preparations that contained from 30 to 100 mM Na⁺ compared with control specimens that contained 20 mM Na⁺. Saadouni et al. (1994) indicated that cocaine binding to rat striatum increased in membrane preparations that contained 10 mM Na⁺ compared with specimens that contained 3 mM Na⁺. This result was attributed to cocaine binding to a Na⁺ carrier complex that facilitated the uptake of cocaine into neurons. However, the affinity of striatal tissue for cocaine decreased when Na⁺ concentrations in membrane preparations exceeded 30 mM Na⁺. Saadouni et al. (1994) indicated that decreased cocaine affinity for binding sites was related to competitive inhibition of cocaine binding by high Na⁺ concentrations. For hair, Kidwell and Blank (1995) described a decrease in cocaine binding due to the presence of NaCl in suspensions. Hair suspensions prepared in 500 mM NaCl bound 5-fold less cocaine than suspensions prepared without NaCl. They indicated that sodium ions and cocaine compete for the same binding sites in hair. These findings may be due to competition between cocaine and sodium cations for melanin binding sites in hair, because similar competition between ions and chemicals occurs for binding sites in synthetic melanin (Larsson and Tjalve, 1978; 1979) and natural melanin (Potts and Au, 1976).

Synthetic melanin has often been used as a model to study how chemicals may bind to natural melanin in tissues. Shimada et al. (1976) examined the binding of cocaine and other drugs to synthetic levodopa melanin. Binding was modeled by using a Type 1 Langmuir Isotherm and was described as reversible drug adsorption to a solid. This approach demonstrated that (±)-cocaine had an affinity (K_d = 1124 nM) similar to that observed for low-affinity cocaine binding to some hair specimens in the present study. Shimada et al. (1976) also demonstrated that synthetic melanin exhibited different affinities for various drugs, including amphetamine (K_d = 9 µM), octopamine (K_d = 3.8 µM) and atropine (K_d = 40 µM). Baweja et al. (1977) used a similar approach and described the affinity of drugs for synthetic melanin by determining the potency of drugs at inhibiting the binding of radiolabeled cocaine. Cocaine was selected because it is reported to have higher affinity for synthetic melanin than do other drugs. Competition between cocaine and other drugs was assessed using a double reciprocal plot of bound cocaine vs. free cocaine to compare the affinity of synthetic melanin for phenothiazines, sympathomimetic amines and other drugs. The investigators determined that binding of phenothiazines and chloroquine to synthetic melanin was greater than that of cyclopentolate, tropicamide and sympathomimetic amines. Nakahara et al. (1995) indicated that the affinity of melanin varies considerably for different drugs. They incubated cocaine, amphetamine and 18 other drugs separately in synthetic melanin suspensions for 2 h and then collected the filtrate after centrifugation. Drug levels in the filtrate were used to determine drug affinity for melanin. Cocaine had the highest affinity of all the drugs, including benzoylecgonine and 11-nortetrahydrocannabinol-9-carboxylic acid.

There are many reports that natural melanin binds chemicals including amphetamine (Harrison et al., 1974a; 1974b) and cocaine (Reid et al., 1994). Potts and Au (1976) described the affinity of melanin for inorganic ions including Li, Na, K and Rb. They reported that affinity increased with atomic weight and was a function of carboxyl groups in melanin. The binding of cocaine to pigmented and nonpigmented irides was described by Patil (1972). The accumulation of cocaine was 18-fold greater in pigmented irides than in nonpigmented irides incubated in identical cocaine solutions. Additionally, the rate of disappearance of cocaine from irides was measured by transferring irides containing bound cocaine into fresh Krebs' solution. Cocaine was washed from the nonpigmented irides at an exponential rate with a disappearance half-life of 10 min. Cocaine bound to pigmented irides was resistant to washing, which suggested that melanin has an affinity for cocaine. Atlasik et al. (1980) examined the in vitro binding of drugs to ocular melanin. Procaine and pantocaine had a strong affinity for melanin compared with other drugs including cloxacillin and fluorescein, which did not bind to melanin. Furthermore, Atlasik et al. (1980) examined the recovery of drugs from melanin by using phosphate buffer, ethanol and other solvent washes. They indicated that binding was reversible and did not involve the formation of a covalent bond, because most bound drug could be washed from melanin by solvents. Potts (1964) also concluded that aromatic compounds do not bind to melanin by covalent bonds and that melanin binding probably involves charge transfer complexes for some compounds.

The mechanisms and structural requirements for the binding of drugs to melanin have not been clearly established. Certain drugs, such as methamphetamine, bind similarly to dark and white hair, which suggests that not all drugs bind melanin in hair (Ishiyama et al., 1983). Nakahara et al. (1995) suggested that the lipophilicity and melanin affinity of a drug determine the rate of its incorporation into hair. They also reported that the extent of incorporation of basic and lipophilic drugs such as cocaine into rat hair after drug administration is generally greater than that of neutral and acidic drugs. The investigators suggested that the pH gradient that exists between blood at pH 7.4 and the acidic hair matrix affects membrane permeability and the disposition of drugs in hair. This gradient may facilitate diffusion of drugs across the membrane between the hair root and blood. Acidic proteins and melanin may bind basic drugs that have diffused into the hair matrix. Similarly, Ward and Lundgren (1954) reported that basic dyes are incorporated into wool fibers to a greater extent than acidic dyes. Melanin is reported to act as a weak cation exchanger that could bind basic drugs such as cocaine in hair (Larsson and Tjalve, 1978; 1979; Potts and Au, 1976).

Specific binding of cocaine to dark hair was consistently greater than binding to light hair in the present study. Joseph et al. (1996) also reported significantly (P < .01) greater in vitro binding of cocaine to dark hair (black and brown) homogenates compared with light hair (blond) homogenates. It was further demonstrated that oxidative degradation of melanin in dark hair by bleaching resulted in decreased cocaine binding. Residual cocaine binding after the bleaching of dark hair was similar to that observed for binding to untreated light hair. Gygi et al. (1996) evaluated morphine and codeine incorporation into rat hair after i.p. codeine administration. They collected dark and white hair from the Long-Evans rats and analyzed both hair types separately. Codeine and morphine levels in dark hair were more than 30-fold greater than those in white hair from the same rat. The investigators also examined the in vitro binding of radiolabeled codeine and morphine to rat hair homogenates. They reported that morphine and codeine binding in vitro to black hair collected from Long-Evans rats was more than 30-fold greater than that to white hair collected from the same rats. Brown hair collected from Dark-Agouti rats bound less morphine and codeine in vitro than black hair from Long-Evans rats, but in vitro drug binding in brown rat hair was 7-fold greater than that in white hair collected from Sprague-Dawley and from Long-Evans rats. The investigators suggested that differences in melanin concentration between hair types were responsible for greater in vivo and in vitro opioid binding to dark hair than to light hair. Gerstenberg et al. (1995) studied the incorporation of nicotine and cotinine into pigmented hair from Brown Norway rats and into nonpigmented hair from Sprague-Dawley rats after s.c. administration of identical doses of nicotine to both strains of rats. The investigators reported that nicotine levels in pigmented hair were 20-fold greater than those in nonpigmented hair. Green and Wilson (1996) examined the incorporation of methadone into dark and white hair collected separately from the same rats after methadone administration. They reported that incorporation of methadone into dark hair of rats was 20-fold greater than its incorporation into white hair, which contained less melanin.

There are limited studies that have investigated whether in vivo differences occur in drug incorporation into dark and light human hair. Kidwell (1992) reported greater incorporation of cocaine into dark hair than into brown hair collected from drug users with similar reports of drug use. Reuschel et al. (1991) collected 48 hair specimens from jail detainees and analyzed specimens for benzoylecgonine, a cocaine metabolite. The concentration of benzoylecgonine in dark hair was 5.6 ng/mg compared with 1.4 ng/mg in light hair. However, a history of drug use before arrest was not obtained, and differences in drug levels between hair types may have been due to differences in the extent and frequency of cocaine use.

In the present study, the differences in specific cocaine binding between dark and light hair appeared to be due to the greater concentration of melanin in dark hair than in light hair. Specific binding to light Caucasoid hair specimens was likely to be less than nonspecific binding, because little melanin is present in light hair. The B_max estimates for high-affinity sites were from 13- to 43-fold greater in dark hair than in light hair. The B_max estimates for the low-affinity sites were from 5- to 18-fold greater in dark hair than in light hair. The greater density of binding sites in dark hair than in light hair may explain why cocaine binding was correlated with hair color. Differences in the affinity of cocaine for dark and light hair do not appear to be responsible for greater radioligand binding by dark hair, because affinity measures for light hair were similar to or greater than those obtained for dark hair in kinetic and Scatchard analyses. Furthermore, the presence of melanin in both hair types may explain stereoselective cocaine binding and the similarities between the potency of cocaine analogs and mazindol at sites in light and dark hair. There were interindividual differences in dissociation constants (table 2), and cocaine binding also varied considerably between specimens of the same color, a result similar to the findings of Joseph et al. (1996). In particular, specific and nonspecific binding was highly variable for dark hair specimens. These findings may be due to differences in the type of melanin in hair specimens or due to differences in melanin concentration in hair specimens that nevertheless look alike in color (Thody et al., 1991).

There were in vitro differences in cocaine binding among ethnic hair types in the present study. Africoid male hair bound more cocaine than all other hair types. In other studies of cocaine incorporation into ethnic hair types, Henderson et al. (1996) reported higher levels of deuterated cocaine in hair of African-American, Hispanic, and Asian subjects than to Caucasians who received identical doses of drug. Cone et al. (1993) reported greater incorporation of cocaine into Africoid head and arm hair specimens than into Caucasoid hair specimens collected from drug users participating in an outpatient study. The investigators indicated that these findings may have been due to ethnic differences in drug incorporation or to differences in drug use between subjects. However, Cone et al. (1993) also reported that benzoylecgonine, morphine and monoacetylmorphine concentrations in Africoid and Caucasoid hair were not significantly different. Mieczkowski and Newel (1993) tested hair and urine specimens collected from 1224 arrestees. They reported that Africoids tested positive at a higher rate for cocaine use by urine and hair analysis than Caucasoids, but the ratio of positive hair tests to positive urine tests was similar for Africoids and Caucasoid arrestees. These findings were attributed to greater cocaine use by Africoids than by Caucasoids as determined by self-reports of drug use. Knight et al. (1996) investigated the disposition of cotinine, a nicotine metabolite, in the hair and urine of 136 Africoid and Caucasoid children. The children were nonsmokers who were passively exposed to cigarette smoke from parents. The level of cotinine in the hair of Africoid children was 2-fold greater than that for Caucasoid children. Additionally, the hair-to-urine cotinine ratio was 2-fold greater for Africoid children than for Caucasoid children. However, differences in cotinine level were not observed in dark hair compared with light hair of Caucasoid children. The investigators indicated that these findings may be due to differences in the amount of melanin in Africoid compared with Caucasoid hair, due to more extensive systemic exposure of Africoid children to cigarette smoke or due to pharmacokinetic differences in nicotine metabolism between the two ethnic groups. In the present study, in vitro differences between ethnic groups in cocaine binding to hair appeared to be related to the melanin content of hair, because there is strong selection for dark hair among Africoids (Wassermann, 1974). Sex also affected cocaine binding to hair; Africoid males bound more cocaine than Africoid females. These findings were previously reported by Joseph et al. (1996), who indicated that differences in the melanin content of hair between Africoid males and females may be responsible for differences in cocaine binding.

Cocaine is incorporated into many different hair types and remains bound to sites in hair for months to years after cocaine exposure has occurred (Cartmell et al., 1991; Ferko et al., 1992; Henderson et al., 1996). In this study, we identified cocaine binding sites in hair by using techniques typically employed in studies of drug binding to receptors in biological tissues. Cocaine binding sites in hair appear to be structurally distinct from binding sites in other tissues, judging by differences in the potencies of various ligands. Binding sites in hair were stereoselective, but the selectivity was considerably less compared with cocaine binding sites in the CNS (i.e., dopamine transporter). Melanin in hair was the probable cocaine binding site and may explain the effect of hair color on cocaine binding in the present study. The potential for greater accumulation of cocaine in dark hair than in light hair is of concern for individuals undergoing drug testing by hair analysis. This study demonstrated a more than 5-fold greater density of binding sites in dark hair than in light hair. This difference could result in greater in vivo cocaine incorporation into dark hair. There also may be differences in the melanin concentration in particular ethnic hair types (e.g., male Africoid hair), that may be responsible for greater cocaine binding compared to other hair types. In vivo studies and further characterization of drug binding sites in hair are needed to determine whether selective accumulation of cocaine and other drugs predisposes certain ethnic groups or individuals with dark hair to test positive for drug use more often than those with light hair.