Roles of drug basicity, melanin binding, and cellular transport in drug incorporation into hair

Hair has become a widely accepted alternative matrix for forensic drug testing. This project examined the roles of drug basicity, drug-melanin binding, and cellular transport of drugs in the phenomenon of preferential incorporation of drugs into darker versus lighter colored hair. Validated assays were developed then used to profile the melanin content in human hair of various colors. Melanin content was then correlated with codeine incorporation into the analyzed hair. Black hair from rats dosed with the basic drug amphetamine was found to contain three times the concentration of amphetamine than white hair from the same rats. In contrast, no difference in N acetylamphetamine (N-AcAp) content was found between black hair and white hair from rats dosed with N-AcAp, a nonbasic amphetamine analog. Cocaine and amphetamine, two drugs that show a hair color bias, bound to eumelanins and mixed eu-/pheomelanins to varying degrees, but not to pure pheomelanin. Benzoylecgonine (BE) and N AcAp, drugs that do not show a hair color bias, did not to bind to any subtype of melanin. Pigmented melanocytes (PM) took up large amounts of the basic drugs amphetamine and cocaine (levels of uptake dependent on melanin content), while keratinocytes and non-pigmented melanocytes (NPM) took up only small amounts of amphetamine. None of the studied cells took up N-AcAp above background levels. While keratinocytes and NPM quickly effluxed most of an influxed basic drug, PM were slow to efflux and only partially effluxed the drug, if efflux media was not refreshed. BE was quickly effluxed from both PM and NPM. Cultured cells influxed amphetamine and cocaine to far greater extents than N-AcAp and BE. This is in accord with the fact that the <italic>non-plasma-protein-bound

ROLES OF DRUG BASICITY, MELANIN BINDING, AND CELLULAR TRANSPORT IN DRUG INCORPORATION INTO HAIR by Chad Randolph Borges A dissertation submitted to the faculty of The University of Utah in partial fulfillment of the requirements for the degree of Doctor of Philosophy Department of Pharmacology and Toxicology University of Utah December 2001 Copyright © Chad Randolph Borges 2001 All Rights Reserved THE UNIVERSITY OF UTAH GRADUATE SCHOOL SUPERVISORY COMMITTEE APPROVAL of a dissertation submitted by Chad R. Borges This dissertation has been read by each member of the following supervisory committtee and by majority vote has been found to be satisfactory. ~v t' I 1-, 2xJCI/ Diana G. ' lIklns ~~ 1/.,;)00 / THE UNIVERSITY OF UTAH GRADUATE SCHOOL FINAL READING APPROVAL To the Graduate Council of the University of Utah: I have read the dissertation of Chad R. Borges in its final form and have found that (1) its format, citations, and bibliographic sty Ie are consistent and acceptable: (2) its illustrative materials including figures, tables, and charts are in place; and (3) the final manuscript is satisfactory to the supervisory committee and is r~adY for submission to The Gra~5: ~ c Av~';,.t 2-1 '-001 ~k.<?, ~ f Douglas E. Rollins Date Chair, Supervisory Committee Approved for the Major Department Chair/Dean Approved for the Graduate Council ~-=-\ '5. c12t--· David S. Chapman Dean of the Graduate School ABSTRi\CT Hair has become a widely accepted alternative matrix for forensic drug testing. This project examined the roles of drug basicity, drug-melanin binding, and cellular transport of drugs in the phenomenon of preferential incorporation of drugs into darker versus lighter colored hair. Validated assays were developed then used to profile the melanin content in human hair of various colors. Melanin content was then correlated with codeine incorporation into the analyzed hair. Black hair from rats dosed with the basic drug amphetamine was found to contain three times the concentration of amphetamine than white hair from the same rats. In contrast, no difference in N-acetylamphetamine (N-AcAp) content was found between black hair and white hair from rats dosed with N-AcAp, a nonbasic amphetamine analog. Cocaine and amphetamine, two drugs that show a hair color bias, bound to eumelanins and mixed eu-/pheomelanins to varying degrees, but not to pure pheomelanin. Benzoylecgonine (BE) and N-AcAp, drugs that do not show a hair color bias, did not to bind to any subtype of melanin. Pigmented melanocytes (PM) took up large amounts of the basic drugs amphetamine and cocaine (levels of uptake dependent on melanin content), while keratinocytes and non-pigmented melanocytes (NPM) took up only small amounts of amphetamine. None of the studied cells took up N-AcAp above background levels. While keratinocytes and NPM quickly effluxed most of an influxed basic drug, PM were slow to efflux and only partially effluxed the drug, if efflux media was not refreshed. BE was quickly effluxed from both PM and NPM. Cultured cells influxed amphetamine and cocaine to far greater extents than N-AcAp and This is in accord with the fact that the non-plasma-protein-bound AUCs of BE and N-AcAp are much gre~ter than cocaine and amphetamine, yet cocaine and amphetamine incorporate into hair to far greater extents than do N-AcAp and BE, regardless of hair color. In conclusion, the data presented in this dissertation demonstrate that amphetamine and cocaine exhibit preferential hair color incorporation (unlike their net neutral analogs N-AcAp and BE) and do so through nondiffusion mediated cellular uptake and subsequent retention via eumelanin binding. v T ABLE OF CONTENTS ABSTRACT ........................................................................................ iv .l·\CKNOWLEDGMENTS ......................................................................... viii Chapter 1. OVERVlEW .................................................................................... 1 Introduction. . . . . . . . . . . . . . . . . .... . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . ... 1 History of Hair Testing ................................................................ 1 Anatomy and Physiology of Hair .................................................. , 4 Routes of Drug Incorporation into Hair.. .. .. .. . . . .. . .. .. .. . .. . .. . .. . .. .. .. . . . ... ... 5 Chemistry, Anatomy, and Physiology of Melanins .............................. 12 Modem Hair Testing... .. .... ...... .. . ...... . .. ......... ...... ......... . . ...... ...... 15 2. RELATIONSHIP OF MELANIN DEGRADATION PRODUCTS TO ACTUAL MELANIN CONTENT: APPLICATION TO HUMAN HAIR ...................... 22 Materials and Methods. . .. .. . .. .. . .. . . . . . .. . . . .. .. . .. .. . . . . .. .. . .. . .. . . . . . . . . .. ... ... 28 Results ................................................................................. 31 Discussion ............................................................................. 53 3. AMPHETAMINE AND N-ACETYLAMPHETAMINE INCORPORATION INTO HAIR: AN INVESTIGATION OF THE POTENTIAL ROLE OF DRUG BASICITY IN HAIR COLOR BIAS .............. , ..... . .. . . . .... . .. . . . . .. . ... .. . . . . ... 58 Materials and Methods ............................................................... 59 Results ................................................................................. 63 Discussion .............................. '" ............................................ 72 4. COCAINE, BENZOYLECGONINE, AMPHETAMINE, AND N-ACETYLAMPHETAMINE BINDING TO MELANIN SUBTYPES ............... 76 Materials and Methods. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . .. 80 Results ................................................................................. 86 Discussion .............................................................................. 99 5. INFLUX AND EFFLUX OF AMPHETAMINE, N-ACETYLAMPHETAMINE, COCAINE, AND BENZOYLECGONINE IN KERATLNOCYTES, PIGMENTED MELANOCYTES, AND NON-PIGMENTED MELANOCYTES ....................................................... ] 16 Materials and Methods. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 119 Reslllts ................................................................................ 127 Discussion. . . . . .... . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 159 6. SUMMARY AND SIGNIFICANCE ................................................... 165 Summary .............................................................................. 165 Significance.......................................................................... 166 APPENDIX ...................................................................................... 169 REFERENCES... . . . . . . ... . .. . ..... .. . ......... . .... . ... . .. ...... ... .. . . . . .. ... . .. . ... ..... . ... 192 vii ACKNOWLEDG!vlENTS I would like to extend much thanks and appreciation to my advisors, Dr. Douglas E. Rollins and Dr. Diana G. Wilkins, for their guidance and counsel in my graduate research and study. Their instruction has been of great value in teaching me the ways of the anal yti cal laboratory. I would also like to thank the other members of my dissertation committee: Dr. Garold S. Yost, Dr. Jeanette C. Roberts, and Dr. Laurence J. Meyer. I have had the pleasure of working in each of their laboratories for a significant period of time during my graduate training. They have each brought a unique and valuable perspective to this project. The rest of the staff at the Center for Human Toxicology has been a great group of people to work with. In particular, I am grateful to Dr. Matthew H. Slawson and Dr. Deanna Hubbard who have provided me with hands-on training for some of the various aspects of this project. In addition, I would like to thank all the additional professors and students here at the University of Utah and at my undergraduate alma matter, Walla Walla College, who have taught me many things and helped shape my scientific philosophy. The concluding research on this project wouid not even have been possible without the valuable cell culture donations of Dr. Dorothy Bennett and Mr. Simon Hill of St. George~s Hospital.rvledical School, London, UK and Dr. Gerald Krueger and Ms. Cynthia Jorgensen of the University of Utah. To them I am greatly indebted. I also express my thanks to the National Institutes of Health and the University of Utah Graduate School, which have provided the financial support for my graduate education. Finally, I would .like to express my heartfelt thanks to my family-to my wife, Laurie, who has been my number one fan and supporter, to my parents, James and Joanne Borges, for their unending encouragement and love, and to my extended family, for their constant support and enthusiasm. This acknowledgments section would not be complete without an expression of gratitude to my Creator, the Alpha the Omega, "Trust in the Lord with all thine heart; and lean not unto thine own understanding. In all thy ways acknowledge Him, and He shall direct thy paths." Proverbs 3:5, 6. ix CHAPTER 1 OVERVIE\V Introduction Hair analysis for xenobiotics began in 1858 with efforts to determine arsenic levels in the hair of a corpse (1). Modern hair testing took another century to begin development. The know ledge about and technology for hair testing has expanded dramatically since the mid 1950s. Today, however, there are still many poorly understood aspects of hair testing, particularly in regard to the fundamental chemistry and biology of how different drugs with their unique and distinctive chemical features incorporate into hair. Ideally, hair testing would be a convenient, non-invasive technique with a long window of detection for monitoring all drugs of abuse. While many drugs are currently tested for in hair, our fundamental knowledge regarding the characteristics and mechanisms for the incorporation of certain drugs into certain hair types is incomplete. What roles, for example, do external exposure, hair treatments, and even hair color play in the incorporation of each tested drug in hair? Fair and accurate hair testing relies upon our knowledge and understanding of these factors. As such, the research presented herein seeks to more fully investigate the roles of hair color (pigmentation), and hair cell drug transport in drug incorporation into hair. History of Hair Testing The history of hair analysis for xenobiotics began in 1858 with the publication of Casper's 'Praktisches Handbuch der Gerichtlichen Medizin' (1) , which translated means, 'Practical Guide to Legal Medicine'. The account describes the determination of arsenic in the hair of a body exhumed 11 years after buriaL The modern saga of hair analysis begins in 1954, with the publication of research conducted by Goldblum et al. describing the detection of barbiturates in guinea pig hair via ultravio.let spectrophotometry (after sample extraction) (2). Twenty years later Harrison et al. (3) described the incorporation of 14C radiolabeled D-amphetamine into guinea pig hair. In a 1979 review of the potentia] of hair as a diagnostic tool, Maugh (4) described efforts to use hair as both a forensic tool and matrix for medical diagnoses. Original efforts to employ hair as a forensic tool sought to use trace element profiles from hair to link suspect(s) to the scene of a crime. Such efforts were soon shown to be futile due to the fact that trace element profiles vary with time and location of head hair. Evidence of severe exposure to trace elements such as lead, arsenic, cadmium, and mercury (4) was, however, obtainable. Ideas to develop elemental analysis of hair as a diagnostic tool were fairly widespread at the time. Efforts in this arena attempted to diagnose disorders such as cystic fibrosis, celiac disease, phenylketonuria, protein-calorie malnutrition (Kwashiorkor), zinc deficiency (ironically, severe zinc deficiency causes hair loss), iron deficiency, juvenile onset diabetes, and learning deficiencies. These efforts, however, did not make a significant contribution to modern medicine. Perhaps due to the excitement surrounding mineral analysis in hair, laboratories began to spring up that would offer nutritional consulting through hair analysis. Surprisingly, this practice is still under debate (5-9). 3 Early hair analysis for trace elements was accomplished through the use of techniques such as neutron activation analysis, photon activation analysis. atomic absorption spectroscopy, and particle-induced x-ray emission. Detection of foreign nonmetallic (i.e., organic or drug) substances in hair was, at the ~ime, essentially precluded due to the lack of techniques sensitive enough to quantitate the low levels of drugs present in hair samples from donors who had not been given repeated, near lethal doses of the drug in question. Research into hair analysis for drugs continued, and in 1979 a seminal paper by Baumgartner et al. (10) demonstrated the detection of opiates in hair through the use of radioimmunoassay (RIA). In addition to their novel techniques, Baumgartner and co­workers succeeded in demonstrating the presence of opiates in hair three months after drug administration as well as time dependent drug deposition along the hair shaft. Soon the field began to grow with regular publications beginning in the early 1980s 01-15). Initial applications of hair testing focused primarily on opiates, phencyclidine, phenobarbital, and cocaine. This was probably due to the availability of antibodies against these drugs. The next major advancement involved the development of electron impact ionization mass spectrometric (EI-MS) methods for the identification of methamphetamine in hair (16). The use of this technology was soon extended to the analysis of tricyclic antidepressants (17), nicotine (17), chloroquine (18) and monodesethylchloroquine (8). Interestingly, although it was not a goal of their investigation, the study by Ishiyama et al. (7) provided the first evidence of a potential hair color bias in drug incorporation into hair. Their results indicate an approximate four fold greater concentration of methamphetamine in the hair of C57 black mice versus the hair of ddY white mice. The only comment made by the authors in regard to this 4 occurrence stated that Harrison's (3) data (which suggested that amphetamine might be incorporating into or binding to melanin in hair) was not supported by their data because they observed methamphetamine in white hair as well as black hair. Thus they suggested that drugs probably bind to protein rather than melanin. An additional improvenlent to the previous EI-MS methods was provided by Suzuki et al. (19) in applying chemical ionization techniques for mass spectrometric analysis of amphetamines in hair. This resulted in significantly better sensitivity and paved the way to modern techniques for drug detection in hair. Anatomy and Physiology of Hair In humans, hair formation first begins primarily on the head when an embryo is approximately 60 days old with the formation of a rudimentary dermal papilla (20). The dermal papilla is a group of specialized undifferentiated fibroblast cells, derived from the mesoderm, that initially form as a clump of cells in the dermis just below the epidermis. The entire hair follicle including associated modified structures such as sebaceous glands, are ultimately derived from the dermal papilla. After the initial cellular organization of the dermal papilla, a peg of cells begins growing into the dermis, pushing the dermal papilla deeper into the dermis and eventually forming an involution. At this stage, the forming hair shaft begins to differentiate, initiating anagen, the first phase of the hair growth cycle, discussed below. During this phase, undifferentiated keratogenic epidermal matrix cells from the dermal papilla begi~ to move upward. Their anatomical fate as part of the hair structure is determined by their initial position relative to the dermal papilla, which remains at the base of the hair follicle (21-24). As cells migrate upward, the hair shaft begins to form from the middle matrix cells. These cells produce large amounts of keratin proteins and eventually harden in a process known as keratinization, which, chemically, is the result of crosslinking of sulfhydryl groups of keratin proteins to form disulfide bonds. Anatomy of the fully formed hair bulb and hair follicle with surrounding structures can be seen in Figures 1.1 (adapted from Robbins (25)) and 1.2, respectively. The ultimate chemical composition of hair varies with location on the body and between individuals, however, it may be of interest in relationship to proceeding sections, that hair consists of protein (65-95%), water (15- 35%), lipids (1-9%), melanin (0.2-1.5%) (see Chapter 2), and trace amounts of minerals (0.25-0.95%) (26). 5 Hair follicles do not continue to grow forever. Rather, after a certain period of growth (anagen), hair follicles undergo a transitional phase, known as catagen, during which cell division halts, the base of the hair shaft keratinizes, and the bulb begins to degenerate. At this point, the hair follicle enters a resting phase, known as telogen, where no cellular proliferation occurs and the now fully keratinized hair shaft can easily be removed from the follicular canal. An increase in metabolic and proliferative activity of the matrix cells reinitiates anagen. This increase in activity is now thought to be due to the divisjon and migration of stem cells from the bulge of the outer root sheath located near the connection point of the arrector pili muscle (27). Figure 1.3 (adapted from Paus and Cotsarelis (28)) demonstrates the relative amount of time spent in each phase of the hair growth cycle. Routes of Drug Incorporation into Hair Drugs can be deposited into and onto hair via a number of routes including through systemic circulation, perspiration, sebum, skin, and the external environment ,..-------14 ;-"---13 ...---1'"""7'-----12 8--------------~ --== --+1---11 Figure 1.1 Hair follicle anatomy. Adapted from Robbins (25) 1) Stratum corneum 2) Epidermis 3) Arrector pili muscle 4) Bulge 5) Dermis 6) Melanocyte 7) Dermal papilla 8) Arteriole 9) Outer root sheath 10) Inner root sheath 11) Sebaceous gland 12) Cuticle 13) Cortex 14) Medulla 6 1------4-------~+--+~~--- ----1-----8 \+-------~--------7 ~------~------6 -1--l~--1----5 2-------4-----+~~~~ 4------\-\-\ Figure 1.2 Hair bulb anatomy. 1) Hair shaft 2) Dermal papilla 3) Melanocyte 4) Arteriole 5) Inner root sheath cuticle 6) Inner root sheath (Huxley's layer) 7) Inner root sheath (Henle's layer) 8) Outer root sheath 7 Sebaceous gland ~'7-- Arrector pili muscle Anagen =:t Club hair Telogen Figure 1.3 Stages of the hair growth cycle. Adapted from Paus and Cotsarelis (28). Hair growth begins with anagen (far right) and ends in telogen (lower right). 8 9 (29, 30). Perspiration, sebum, and the external environment can only deposit drug on the outside of the hair follicle. For most hair testing purposes this is an undesirable situation, since drug on the external surface of hair is not indicative of ingestion. In an ideal situation, hair testing would only analyze drug incorporated into hair through systemic circulation. This would allow for the least number of interferences and inconsistencies, e.g., drug from smoke or other residue in the environment, and inconsistencies due to inter-individual differences in perspiration and sebum secretion. This would also eliminate the need for hair washing procedures prior to analysis. The real situation is not ideal, however, as each of the above routes may substantially contribute to drug incorporation into hair. Nevertheless, hair analysis relies on the fact that drugs can be incorporated from systemic circulation into hair follicles. This route for drug incorporation into hair is arguably the most important because it provides the basis for consistent, reliable measurements of drug ingestion, and, at least in rat models, is the route by which the majority of drug is incorporated into hair. (Environmental exposure is easily controlled in animal studies, and, except for the soles of their feet, rats lack sweat glands.) The simplest model for drug incorporation into hair from the bloodstream involves passive diffusion of drug from arterial capillaries to extracellular fluid to the inside of hair forming cells. Once inside the cells, the drug must be retained long enough for the cell to move up and out of the zone of differentiation and biological synthesis. While in certain drug-animal models, drugs may incorporate into hair from the bloodstream in a linear, dose-dependent manner suggestive of passive diffusion down a concentration gradient from blood into hair (31), accumulating evjdence argues that this model is probably not 10 accurate (29). A number of drugs incorporate into hair to a greater extent than their metabolites or related compounds despite the fact that the area under the plasma concentration versus time curve (AUC) is much greater for the compounds that are barely detected in hair. For example, in various animal and human models, tJ.9_ tetrahydrocannabinol (THC) (32-35), cocaine (29, 36-39), nicotine (40), amphetamine (41,42) and 6-acetylmorphine (43), all are found in greater concentrations in hair than Il-nor-9-carboxy-tJ.9 -tetrahydrocannabinol (THC-COOH), BE and ecgonine methyl ester, cotinine, N-AcA.p, and morphine, respectively, while the AUCs or plasma concentrations of the metabolites (or related compounds) are much greater than the plasma concentrations of the parent compound. These data suggest that factors other than passive diffusion down a concentration gradient or even simple selective diffusion through a semipermeable membrane may be important. Factors such as selective . . transport of certain drug molecules into hair forming cells and binding of drugs by hair components (i.e., protein, melanin, and possibly lipids) may play major roles in determining the amount of a given drug incorporated into hair. Although little, if any, ~. research has been conducted to determine whether hair cells may take up drugs by mechanisms other than passive diffusion through the cell membrane, a fair amount of research has been done to characterize the binding of drugs to hair components (i.e., protein (44-49), melanin (50-62), and lipids (63)). A major controversy still exists in this area, especially in relation to the ability of drugs to bind to melanin, and whether or not this can produce a hair color bias (40, 63-74). Nakahara (75) (and others) have carried out a large amount of research to determine what physicochemical factors are most important for a drug to be incorporated into hair. Findings indicate that the greater the melanin affinity (40,43,63-71), lipophilicity (41, 75), and basicity (41, 42, 75) of a drug, the more it incorporates into hair. 11 Melanin affinity is important because it represents a mechanism whereby significant amounts of drug can bind inside hair cells and be retained as the cells keratinize and move up into the hair shaft. A large number of drugs have .been shown to bind to melanin, including amphetamine (55, 56, 76), chloroquine (51, 53, 57), chlorpromazine (47,48,51,54), cocaine (55,56,76), phencyclidine, methylenedioxymethamphetamine (43), tricyclic antidepressants (56), 1-methyl-4- phenylpyridinium (MPP+) (77,78), clenbuterol, nortestosterone, diethylstilbesterol (54), paraquat (50,51), and streptomycin (79) to name a few. Even some cationic metals such as iron, manganese, copper, lead, nickel, magnesium, zinc, cadmium, aluminum, scandium, lanthanum, and indium (51, 80-83) are documented to bind to melanin. The exact chemical nature of drug-melanin binding is not well understood, but it is thought to involve various types of binding interactions between drugs and melanin orthoquinones, phenolic groups, carboxylic acids, indole-amines, and/or van der Waals interactions between stacked indole units (50, 51, 53, 83). Because of their affinity for certain types of drugs (e.g., basic drugs), melanins appear to playa major role in the incorporation of such drugs into pigmented hair (40,63-71). Interestingly, drugs can also bind to' melanins in locations throughout the body other thaJ? hair. In fact, many of these cases have implicated toxicological effects (58). For example, drug binding to the uveal tract of the eye (84,85), ear (cochlear) (86, 87), and neuromelanin (77,78,88, 89) has been implicated in the negative toxic effects of the drugs to surrounding tissues. Lipophilicity 12 is important because it is thought to allow permeation through biomembranes such as in capillary beds and hair cell membranes. Interestingly, Nakahara (43) showed that for 19 drugs, when melanin affinity (Ka) is multiplied by lipophilicity (log P, see Kaliszan et al. (90)) and correlated to the hair incorporation ratio ((ICR) = [Drug in Hair]lPlasma AUC), an R2 value of 0.979 is obtained. Basicity is probably related to melanin affinity, i.e. positively charged, basic drugs bind best to negatively charged, acidic melanin. The isoelectric point of hair is approximately 3.7 (91) and is probably mostly derived from the acidity of melanin, however, acidic sites on hair keratin cannot be ruled out as drug binding sites (45). Data from Dehn et al. (92) suggest that drugs such as nicotine and cotinine may be covalently incorporated into the melanin polymer. While this phenomenon is of interest from a biochemical standpoint, it cannot contribute to drug detection in hair because the drug molecules would have to be cleaved during the extraction process at the exact carbon-carbon juncture where they were initially joined to the melanin polymer. Covalently bound compounds would not be detectable with currently available techniques. Chemistry, Anatomy, and Physiology of Melanins The two critical ingredients for cellular biosynthesis of melanin are the amino acid tyrosine and the rate limiting enzyme tyrosinase. In the Golgi apparatus-derived (93) melanosome, tyrosine undergoes a series of oxidations to produce both the black colored eumelanin and the reddish-brown colored pheomelanin polymers (Figure 1.4). The structures of eumelanin and pheomelanin are not shown because they are varied and not well characterized. Eumelanin is a polymeric combination of carbon-carbon linked indolequinones and carboxylated pyrroles derived primarily from 5,6-dihydroxyindole 13 Figure 1.4 Biosynthetic pathways of (A) eumelanin and (B) pheomelanin. A ---..-. Eurnelanin 02 -------- ~COOH HXJo .p' " H ~ I H 5,6-Dihydroxyindole (DHI) ZC02 o HM-COOH Dopachrome ~P2 t (Dopachrome Tautomerase) I 02 H~ COOH H~r H 5,6-Dihydroxyindole-2-carboxylic acid (DHICA) ---0-2-. .. H ~COOH H~ COOH H~r H Leucodopachrome r O~COOH HO~ NIIz Tyrosinase HAJ NH2 -0.2. . Tyrosinase ~V NH2 Tyrosine DOPA Dopaquinone ~ B ~COOH HO~ MI2 Tyrosine -0.2. . Tyrosinase Pheomelanin nxn:COOH ~ 02 'NH ~ H ~ 2 Tyrosinase DOPA o ~~COOH ~ NH2 Dopaquinone r Cysteine H ~COOH HOY NH2 HM-tN,H2 S ~COOH NH2 5 -S -Cystei n y ldopa ~ H~COOH ~ NH2 HOO~S ! + H ~ C.O OH S ~COOH NH2 2-S-Cysteinyldopa + llMI t~NH 2 HOO~~ COOH Benzothiazine derivatives Ul (DR!) and 5,6-dihydroxyindole-2-carboxylic acid (DRICA), while pheomelanin also includes related benzothiazine units derived primarily from 2-cysteinyl-S-Dopa (2- CysDOPA) and 5-cysteinyl-S-Dopa (5-CysDOPA) (94, 95). 16 Melanocytes derived from the neural crest associate with the matrix cells in a relationship whereby melanin granules, known as melanosomes, are transferred to the keratogenic matrix cells (96-98). This process, which gives hair its color, is thought to occur through both exocytosis of melanosomes and uptake by keratinocytes andlor keratinocyte phagocytosis of invading melanocyte dendrites (93, 98-102). As with most other biological processes, melanin biosynthesis is under the control of honnones, receptors, and enzymes. Pro-opiomelanocortin is produced in the pituitary gland and (as relates to melanogenesis) is cleaved to produce adrenocorticotropin and modified to produce a-melanocyte stimulating honnone (103). These peptide hormones activate G­protein coupled melanocortin receptors (l04) which activate adenylyl cyclase and lead to an increase in intracellular cyclic adenosine monophosphate (cAMP) (103, 105). cAMP activates protein kinase C beta which. in tum, activates tyrosinase through phosphorylation of serine residues in the cytoplasmic domain of the melanosomal­membrane bound protein (105, 106). Melanogenesis is also regulated by light during the tanning process. Light produces two tanning phases-immediate tanning and delayed tanning (107). Melanin formed during immediate tanning appears to be structurally different from constitutive melanin. Although exact structures are not known, it is derived from photo-oxidation polymerization products of melanogenic precursors such as DHI and DRICA. Its fonnation is oxygen dependent (107). Delayed tanning is not oxygen dependent and 17 produces a longer lasting tan due to increased melanin synthesis, melanin transfer to, and distribution in epidermal cells (07). Although the biochemical pathway for delayed tanning is not well understood, studies by Schallreuter and co-workers (104) suggest that UVB can release TNF-a, which induces GTP-cyclohydrolase 1, which, in tum, synthesizes L-erythro-5,6,7,8-tetrahydrobiopterin (6BH4), a cofactor for phenylalanine hydroxylase which converts phenylalanine to tyrosine-the amino acid required for melanin synthesis. 6BH4 is known to inhibit tyrosinase by a noncompetitive allosteric mechanism (108), but UVB light can photo-oxidize 6BH4 to 7,8-dihydroxanthopterin, reactivating tyrosinase and overall enhancing pigmentation (104, 109). Evidence presented by Palumbo et aI. (110), suggests that in addition to its well known copper cofactor, tyrosinase may also require Fe2 + to act as a redox exchanger with the cupric ions at the active site of the enzyme. The summary of scientific literature available on tyrosinase research suggests that it is indeed a complexly regulated enzyme. Other, perhaps indepe'ndent, pathways may also be involved in delayed tanning. Additional data suggest that DNA damage induced by UV light can induce melanin production (111). Interestingly a primary mediator of this response appears to be the thymine dinucleotides produced during DNA excision repair after UV irradiation. Presented to cells alone in the absence of UV light, such dinucleotide dimers can induce melanin production that closely mimics that induced by UV irradiation itself. The biochemical pathways underlying this process, as well as UV light-induced, tanning in general, however, remain to be confirmed and fully elucidated. 18 Modem Hair Testing Applications for Hair Testing Even though all the time course, metabolic, and potential hair color bias effects have not been completely worked out for hair testing, hair testing is now the most widely accepted alternative to urinalysis for drug testing in the United States (112). Among hair testing's applications include preemployment screening, drug recidivism screening for patients in antiabuse programs, athlete testing, school drug testing, and home hair: testing kits (for parents to screen their children). Hair testing has been applied to virtually every drug-testing situation. An interesting proposal for the use of hair testing would be in compliance monitoring programs (113-119). The purposes of such analyses would include monitoring patient compliance of prescribed medications (Le., neuroleptics, antiepileptics, buprenorphine, and methadone) and even potentially using an "inert" drug such as ofloxacin that moves up in the hair shaft with time to track another drug's ingestion during a specified monitoring period. In theory, drugs stay with the hair cells they incorporate into and move along with the hair shaft as it grows. This phenomenon has been documented for ofloxacin (120), rhodamine, and fluorescein (49). Advantages of Hair Testing There are two primary advantages that hair testing provides over traditional urinalysis and plasma testing methods. First, hair testing is less invasive; collecting a few hairs from the scalp takes less effort for both laboratory personnel and the patient than does acquiring a urine or plasma specimen. Integrity of the sample can also be ensured because laboratory personnel (or potential employers) can collect the sample directly from the subject without the difficulties of having to trust the subject or having to observe 19 urination. Second, hair testing provides a larger window of detection than does plasma or urinalysis. A commonly sited time frame for detection of common drugs of abuse is 90 days (112)l although this has not been rigorously tested for every drug tested for in hair. Disadvantages of Hair Testing There are a few difficulties specific to hair testing that are not encountered with traditional plasma and urine testing. First, drugs from the environment (i.e., smoke) can adhere to the outside of hair and produce a positive test result even if no drug has been ingested systemically. This can be a problem with all drugs that are smoked (which, unfortunately, includes all the major drugs of abuse). In theory it may be possible to circumvent this difficulty by applying a proper wash procedure to the hair prior to analysis. The wash procedure must be documented to wash off only external drug contamination (without removing systemically incorporated drug, presumably inside the hair shaft) and to remove all external contamination. Paulsen et al. (121) have demonstrated that several currently employed laboratory hair wash procedures (including methanol, 0.1 M phosphate, pH 6.0 and pH 8.0, and isopropanol and phosphate buffer, pH 5.5) can significantly alter the reported levels of cocaine in hair from systemic incorporation. Unpublished data from the Center fot Human Toxicology at the University of Utah indicate that both rat and human hair exposed to cocaine freebase smoke, even after washing with phosphate buffer, pH 5.5 and methanol, contains enough cocaine to potentially be reported positive for drug abuse-if benzoylecgonine (BE) is not taken into account. If not handled carefully and appropriately, the former phenomenon can lead to false negative results and the latter to false positive results. Therefore special care to properly interpret hair testing results, e.g., employing a BE to 20 cocaine ITlinimum ratio (that is, a metabolite to parent dnlg minimum ratio) for a positive test and performing thorough validation studies, must be taken to produce accurate hair testing reports. In addition to smoke, drugs can also be deposited on the hair from perspiration and sebum. It would not present a problem if this route of incorporation produced a positive test, but interindividual differences in the concentration of drug in perspiration and sebum and the amount of perspiration and sebum secreted are just starting to be investigated (122, 123) and may play significant roles in determining hair test outcomes if wash procedures are routinely employed. A second major difficulty involving hair testing is the effect of hair treatments such as bleaching and dyeing on ~air test outcomes. Included in this category is head shaving, which, if head hair is the only validated hair type for drug testing, could easily provide a means of evading hair testing. The effects of hair treatments such as bleaching and dyeing on hair drug test outcomes are not well investigated. However, studies by Kidwell and DeLauder (124, 125) suggest that cosmetic hair treatments may alter the physical properties of hair, thereby altering its drug binding capacity and increasing its accessibility to external contamination; thus potentially altering a hair test outcome. A third problem plaguing hair testing and a major focus of this dissertation is the potential for hair color bias; that is, the phenomenon where more drug is incorporated into dark hair than light colored hair-all else held equal. Hair color bias arises for a drug because of its binding affinity for the hair melanin itself. As mentioned above, because of their affinity for certain types of drugs (e.g., basic drugs) (50-62), melanins appear to playa major role in the incorporation of such drugs into pigmented hair (40, 21 63-71). The existence of a hair color bias means that for hair testing to be fair to all members of society, hair testing for drugs that demonstrate a hair color bias must include normalization to the amount of melanin in hair-especially, and perhaps specifically, the amount of eumelanin in hair. Studies at the Center for Human Toxicology at the University of Utah (126) have shown that incorporation of codeine into human hair is highly correlated with the amount of eumelanin in hair while the amount of pheomelanin does not seem to matter much. This is in agreement with studies to be presented later in this dissertation showing that pure eumelanins bind basic drugs but that pure pheomelanins do not bind basic drugs. In summary, hair testing provides advantages over traditional drug testing techniques that make it a very attractive alternative matrix for drug detection. As hair testing technology becomes more popular, however, we must make sure not to extend hair testing beyond our fundamental knowledge of the chemical and biological processes involved in drug incorporation into hair. Only in this manner can society be assured of fair and accurate drug testing outcomes from this innovative technique. Research Objectives The overall objectives of this research were to assess the ability of basic and non­basic related drugs to incorporate into black and white hair, bind to melanin, and move into and out of hair cells. The following hypotheses and specific aims were designed to accomplish these goals: 1. Hypothesis: The amount of melanin in hair is directly proportional to the amount of systemically administered codeine that is incorporated into hair. 22 Specific Aim 1 a: Establish subtype specific melanin assays for quantitating the amounts of DRI, DHICA, 2-CysDOPA, and 5-CysDOPA-derived melanin in biological melanin­containing samples. Specific Aim 1 b: Correlate the amount of each melanin sUbtype in hair to the amount of codeine incorporated into the hair through linear regression analysis. 2. Hvpothesis: Amphetamine will show a hair color bias while its non-basic analog N -AcAp will not. Specific Aim 2: Determine the concentrations of amphetamine and N-AcAp in black and white rat hair (from the same animal) after systemic administration of the drugs. 3. Hypothesis: Drugs that demonstrate a hair color bias will bind to DHICA­melanin with the same affinity, but greater capacity than to DHI-melanin, and with lower affinities and capacities for 2-CysDOPA-melanin and 5-CysDOPA-melanin. Analogs of these drugs that do not demonstrate a hair color bias will not bind to any type of melanin. Specific Aim 3: Determine the in vitro binding affinities and capacities of cocaine, BE, amphetamine, and N-AcAp for DHICA-melanin, DHI-melanin, mixed DHII2-CysDOPA­melanin, mixed DHII5-CysDOPA-melanin, and pure 5-CysDOPA-melanin. 4. Hypothesis: Drugs that incorporate into hair to greater extents than their net­neutral congeners will be taken up faster, be effluxed slower, and have higher equilibrium uptake concentrations than their congeners-in keratinocytes, PM, and NPM. In addition, basic drugs will be transported at the same rate into PM and NPM, but will have higher equilibrium concentrations in PM than their nonbasic analogs. Finally, plasma protein binding will not completely account for the drug to metabolite or congener ratio differences seen in plasma and hair. That is, hair cell selectivity (as opposed to free drug concentration in plasma) is suspected as the major factor in determining how much of a drug gets into hair cells. Specific Aim 4a: Profile the influx and efflux time courses of cocaine, BE, amphetamine and N-AcAp in cultured keratinocytes, PM, and NPM, and relate this data to in vivo data for drug incorporation into hair. Specific Aim 4b: Determine the extent of plasma protein binding of cocaine, BE, amphetamine, and N-AcAp to assess the relative amounts of free cocaine vs. free BE in plasma (after cocaine administration), and free amphetamine vs. free N-AcAp (after equi-dosing of either drug). CHAPTER 2 RELATIONSHIP OF MELANIN DEGRADATION PRODUCTS TO ACTUAL MELANIN CONTENT: APPLICATION TO HUMAN HAIR a Numerous studies have confinned that melanin pigments play an important role in the incorporation of drugs into hair (3, 40,63-65,69-71, 127, 128). This may lead to a hair color bias in the interpretation of hair testing results. The purpose of this investigation was to detennine the yield of melanin subtype-specific chemical markers produced from chemical degradations of pure melanin subtypes, then apply this information to profile the melanin content and character of a range of human hair types. Melanins are highly heterogeneous pigment polymers that give hair and skin their color. These pigments are typically divided into two categories: the black eumelanins and the reddish-brown pheomelanins. Eumelanin is composed of the tyrosine-derived indole units DHI and DHICA (129, 130) (Figure 2.1), Pheomelanin is composed of tyrosine and cysteine-derived units, generally thought to be constructed into benzothiazine monomers that make up the pheomelanin polymer (96). 5-CysDOPA and 2-CysDOPA are thought to be the major pheomelanin building blocks (132) (Figure 2.1). It must be kept in mind, however, that in vivo, melanins are generally not homopolymers a Relationship of Melanin Degradation Products to Actual Melanin Content: Application to Human Hair" by Chad R. et al., from Analytical Biochemistry, Volume 290, 116·125, copyright © 2001 by Academic reprinted, with modification, by permission of the publisher. 25 Figure 2.1 Biological synthesis and chemical degradation pathways of A) eumelanin and B) pheomelanin. (Adapted from Ozeki et al. (131) and Kolb et al. (132» A Hooe Hooch N H Pyrrole-2,3-dicarboxylic acid (PDCA) i Alkal~elH202 .. Eumelanin 02 I Alkaline/H202 ~ tJ. TRP] Hooe (DHICA oxidase 'In Hooe-l(N~eooH H Pyrrole-2,3,5-tricarboxylic acid (PTCA) HO~ ~j IIO H 5,6-Dihydroxyindole (OHI) , C02 OX):)-eOOH ~ ~ HO N ~RP2 Dopachrome (Dopachrome Tautomerase) r , 71~ eOOH H0:(Jc}- . ~ N HO H 5,6-Dihydroxyindole-2-carboxylic acid (Ol-I1CA) 02 'TyroSinase ~eOOI'1 HO~ NH2 Tyrosine tv Q\ B ~eOOH 02 HOV NH2 --+- Tyrosine Tyrosinase Ho~eOOH HOY NH2 s \-COOH NH2 5-S-CysteinylDOP A rCysteine + H?O I ~NH 2 HO ~ eOOH S \-COOH NH2 2-S-Cysteiny lDOP A I Pheomelanin I ~ HI Hydrolysis Ho~eOOH + HO~ ~H2 N ~ I NH2 N~eOOH Hooe~S Hooe~ s Benzothiazine derivatives I I I I I I , I I I I I I , ______ H~o~eOOH H2NV NH2 H2N~eOOH HOV NH2 4-Amino-3-hydroxyphenylalanine (ARP) 3-Aminotyrosine (3AT) 27 of a single building block or even made solely from eumelanin or pheomelanin monomers-rather they are mostly ill-defined heteropolymers made up of both eumelanin and pheomelanin building blocks (96). 28 Melanins are biologically synthesized through what is thought to be a free radical­mediated process (133-135) in the melanosomes of melanocytes with the aid of the enzyme tyrosinase. In the cell, melanins are covalently linked to proteins to produce structures referred to as melanoproteins. Once melanoproteins are made they are transferred to keratinocytes (96, 136) where they are effectively displayed on the body's surface. While the biological role of melanin is not completely understood, it is known that eumelanin in dark skinned individuals serves in. a photo-protective manner against damaging ultraviolet light. On the other hand, large amounts of pheomelanin in light skinned/freckled individuals appears to make them more susceptible to skin cancer (137- 139). The natural role of melanin in hair, however, remains unknown. It is not a "natural" role, but recent studies have shown that melanin plays a crucial role in binding many drugs that are incorporated into hair. Interestingly, a number of drugs with a basic nitrogen moiety such as cocaine (63), methadone, (140), codeine (127), phencyclidine (69), haloperidol (70), ofloxacin (71), and nicotine (40) have all been shown to incorporate in greater amounts into hair of darker vs. lighter pigmentation. At the same time the nonbasic, but nitrogen-containing drug phenobarbital has been shown to have no preferential incorporation into hair of darker pigmentation over hair of lighter pigmentation (65). The theme for preferential incorporation into hair of darker pigmentation appears to be a basic nitrogen moiety. This is not surprising when one considers both the relatively high content of negatively charged carboxyl groups on 29 melanin (96, 129) and the fact that substances with cationic properties such as amines and metals are bound to melanin through ionic interactions (58, 83). If the role of melanins in drug uptake into hair is to be fully understood, the chemical nature and quantity of the melanins in hair must be elucidated. Sensitive HPLC methods for characterizing specific melanin subtypes, that is, DHI and DHICA in eumelanin, and 5-CysDOPA and 2-CysDOPA in pheomelanin, have been developed (130, 132). These methods involve chemically degrading eumelanins and pheomelanins with hydrogen peroxide and hydriodic acid, respectively, to produce PDCA, PTCA, AHP, and 3AT (Figure 2.1)-chemical markers for DHI, DHICA, 5-CysDOPA, and 2- CysDOPA, respectively. This chapter reports the yields of PDCA, PTCA, AHP, and 3AT that are produced upon alkaline hydrogen peroxide or hydriodic acid degradation of melanins made solely from monomers of DHI, DHICA, 5-CysDOPA, or 2-CysDOPA, using modified versions of the original degradation methods (130, 132, 141 )" The eumelanin assay was modified to include sodium hydroxide as the alkaline agent. This permits complete degradation of hair samples during the incubation time. Thus the yields for PDCA and PTCA that were previously reported (130) for this assay needed to be re-determined. The previously reported yield for combined 3AT and AHP from 5-CysDOPA melanin was 20% (141). Individual yields from 3AT and AHP from 2-CysDOPA melanin and 5-CysDOPA melanin, respectively, are now reported. To ensure accuracy when analyzing biological samples, cross-reactivity studies were conducted to account for production of chemical degradation markers from unexpected sources. The modified methods and newly determined yields were used to determine the melanin SUbtype composition of a variety of human hair samples. Finally, the melanin content was correlated to codeine incorporation to assess any possible relationship between the two variables. Materials and Methods Materials 30 Tyrosinase, 3AT, hydriodic acid, and hydrogen peroxide were purchased from Sigma Chemical Co. (St. Louis, MO). Hypophosphorous acid (H3P02) was purchased from Aldrich Chemical Co. (Milwaukee, WI). DHI and DHICA were made according to the method ofWakamatsu and Ito (142). 5-CysDOPA and 2-CysDOPA were made according to the method of Ito et aI. (143). PDCA and PTCA were made according to the method of Ito and Wakamatsu (130). AHP was isolated from hydriodic acid hydrolyzed 5-CysDOPA melanin by the solid phase extraction method described below for extraction of AHP and 3AT from hydriodic acid hydrolysates, but using 2 M HCI instead of 0.3 M KCl to elute AHP from the columns. Eumelanins were made according to the method of Ito et aI. (144) and pheomelanins were made according to the method of Ito and Fujita (141). The structure and purity of all synthetic compounds (except melanins) were confirmed by NMR and/or mass spectrometry (see Appendix). IH NMR spectra were obtained on a Bruker AF-200 MHz spectrometer. Mass spectra for synthetic compounds were obtained with a HP 1100 series LCIMSD mass spectrometer in FIA mode equipped with an electro spray ion source. All other chemicals used were of the highest purity available. 31 Melanin Analysis Melanins were prepared for analysis by homogenization in water at a concentration of 2 mg/m1 using a glass homogenizer operated by a drill press. Hair was cut into small « 2 mm) pieces prior to being weighed out. Typically 0.2 mg melanin and S mg hair were used for analysis. Eumelanin analysis. Samples were chemically degraded then analyzed for PDCA and PTCA via the method of Ito and Wakamatsu (130) with modifications: To a given sample (-S mg) in 100 ~l water in a screw-capped tube was added 820 ~l O.S M NaOH, 80 III 3 % H20 2, and 40 nmol phthalic acid as an internal standard. Samples were then heated in a boiling water bath for 20 min. After cooling, 20 ~I 10% Na2S03 and 2S0 ~1 6 M HCI were added. Samples were then extracted twice with 7 m1 ethyl acetate. The ethyl acetate was dried under a stream of air at 45°C and the residue redissolved in 1 ml starting HPLC mobile phase. HPLC analysis was carried out with a Waters 600E multisolvent delivery system equipped with a Waters 600 controller and Waters 717plus autosampler. One hundred microliter samples were injected onto a Phenomenex (Torrance, CA) Luna S Ilm C 18 250 x 4.6 mm column at a temperature of 55°C. Analytes were detected with a Varian 90S0 variable wavelength UV detector set at 280 nm. HPLC mobile phase consisted of 0.01 M potassium phosphate buffer, pH 2.1, and methanol at a flow rate of 0.8 ml/min under the following gradient: 98%/2% Aqueous/organic ramped evenly from time 0 to 14 min to 40%/60% aqueous/organic, held at 40%/60% aqueous/organic for 6 min followed by ramping back to 98%/2% aqueous/organic over S min. Amounts of PDCA and PTCA were quantitated using PDCA:phthalic acid and PTCA:phthalic acid peak height ratios compared to a standard curve made from pure PDCA and PTCA standards subjected to alkaline hydrogen peroxide degradation. Pheomelanin analysis. Samples were analyzed for pheomelanin content according to the method of Kolb et al. (132) with modifications. To a 100 ~l suspension of melanin (0.2 mg) or hair (5 mg) in water placed in a screw-capped tube was added 500 J.ll 57% HI, 20 J.lI 50% H3P02, and 20 nmol L-a-methyIDOPA (L-a-MD) as an internal standard. Samples were capped tightly and hydrolyzed at 130°C in an oil bath for 16 hrs. After cooling, samples were evaporated under reduced pressure in a Savant (Holbrook, NY) Speed-Vac SPD121P concentrator at 55°C. Users of this device should note that HI can corrode the top of the lower magnet assembly and thus eventually compromise the vacuum established in the concentrator. Dried residue was redissolved in 1 rnl 0.05 M lithium phosphate buffer, pH 4.0, but not adjusted to pH 4.0. AHP and 3AT were extracted with aromatic sulfonic acid (SCX) solid phase extraction (SPE) columns (International S0rbent Technology, Mid Glamorgan, U.K.) containing 100 mg sorbent and a 1 ml reservoir volume. SPE columns were washed with 1 ml methanol (2x), and then 1 mllithium phosphate buffer (3x) prior to sample application. Columns were washed with 1 ml water (2x) then eluted with 2 ml 0.3 M KCl, pH 8.5. Twenty microliters of eluant were injected into the same HPLC system used for eumelanin analysis, but equipped with a Waters 464 pulsed electrochemical detector equipped with a glassy carbon electrode set at +400 mV relative to a Ag/AgCl reference electrode. Mobile phase consisted of 99% 0.01 M potassium phosphate buffer, pH 5.7 containing 1 rnM sodium octanesulfonate, and 0.1 mM disodium EDT A/I % methanol at a flow rate of 0.9 ml/min. Amounts of AHP and 3AT were quantitated using AHP:L-a-MD and 33 3AT:L-a-MD peak height ratios compared to a standard curve made from pure AHP and 3AT standards subjected to hydriodic acid hydrolysis. Statistical Analysis Comparisons between groups were made with a two-tailed Student's t-test assuming homogeneity of variances. Results significant at p :::; 0.05 are reported. Results Degradation product yields Representative chromatograms from melanin samples obtained from alkaline hydrogen peroxide degradation and hydriodic acid hydrolysis are shown in Figures 2.2A and B. The small PDCA peak height is explained by the relatively low yield of PDCA. Degradation product yields from their respective pure melanins as well as overall assay precision can be seen in Table 2.1. Degradation product yields are based on a nlass / mass ratio of the amount of degradation product detected / amount melanin degraded. As seen from these data, yields of melanin degradation products are quite consistent. To demonstrate assay linearity with different sample sizes, varied amounts of both hair and melanin were subjected to both eumelanin and pheomelanin analysis. As shown in Figure 2.3 and Table 2.2, increasing amounts of melanin or hair produce linearly increasing amounts of melanin degradation markers. Figure 2.3 demonstrates that as increasing amounts of eumelanin are oxidized, linearly increasing amounts of PDCA and PTCA are produced. Table 2.2 shows that in general, linear relationships are found for plots of the amount of degradation marker produced vs. the amount of pure melanin subtype degraded, and for plots of the amount of melanin subtype found in a hair sample Figure 2.2 Sample chromatogramsfrom melanin chemical degradation: A) Alkaline hydrogen peroxide degradation of 0.2 mg melanin made from 750/0 DRI/ 25% DHICA (w/w). B) Hydriodic acid hydrolysis of 25 J,.lg melanin made from 50% 5-CysDOPA / 50% 2-CysDOPA (w/w) (100 nA full scale). 34 35 A 0.10 - PTCA Phthalic Acid I I I 0.00 15.00 30.00 Minutes B 30.00 - 20.00 - 10.00 - I ~ \ I I \ Me~hYIDOP A O,OOJ-'II~~~ -10.00 - \ I II I 0.00 V I 6.00 Minutes I 12.00 36 Table 2.1 Percent Ine lanins and with hUl1lan hair PDCA Yield PTCA Yield AHP Yield 3AT Yield M,','w, "'~""'''''''''''''~'l)'''''''W~'''''_N' '." Melanins Intra assayC 0.37% ± 0.0070/0 4.5% ± 0.2]% 23% ±2.0% 16% ± 0.52% Inter assayd 0.37% ± 0.025% 4.8% ±0.22% 23% ± 1.2% 16% ±0.87% DHI Melanin DHICA Melanin 5-CysDOP A Melanin 2-CysDOP A Melanin Haire 13000 ± 420 6400 ± 710 400 ± 38 1200 ± 37 Inter assavu , f 14000 ± 1700 6200 ± 90 680 ± 58 1000 ± 45 bExpressed as the amount of degradation product detected / alTIOunt nlelanin degraded (w/w percent yield ± standard deviation). c n=4 d eExpressed as ng melanin per mg hair (± standard deviation). Black hair was used for PDCA and PTCA determination and hair was used for AHP and 3AT determination. fFor pheomelanin data, interassay precision was assessed with hair from a different individual than that used to assess intra­assay precision. vol -J 38 Figure 2.3 Plot of the amount of eumelanin degradation marker produced vs. the amount of eumelanin analyzed. This plot demonstrates the melanin assay linearity found when increasing amounts of melanin or human hair are subjected to analysis. See Table 2.2 for complete data set and statistical analysis. o ".,_~.,,~. ________ ~_w ____ ~. ________ ., _______ ~____ 0 0 C\l c::(c::( U(.) 01- c..c.. • • 0') 0') 0') o II C\J a: ,0... .. ~ 0) 0) . o II C\J a: (6n) pa~npoJd Ja>lJBW UO!IBpBJ6ap U!UBlaWna 10 IUnOWV ~ 0 0 C\J 0 0 39 ..--.... C') ::J ~ -c G) -N- --C- >< 0 .c-: c: -m G) E ::J G) "I- .0. c: ::J 0 E <t Table 2.2 Regression analysis on melanin assay linearity, shown as increasing all10unts of pure melanin subtype or human hair is subjected to analysis. For degradation markers (PDCA, PTCA, 3A1: and AIIP), the data corresponds to a plot of the anlount of degradation lnarker produced (in fig) vs. the alnount of pure lnelanin subtype degraded (in pg). For a melanin subtype (the remaining entries), the data correspond to a plot of the amount of melanin sllbtype found in a sample (in fig) vs. the anlount of hair subjected to analysis (in mg). Errors are expressed in terms of standard deviation as denlonstrated by Anderson (145). R2 Slope Slope error y-intercept y-intercept error PDCA 0.994 0.00297 6.96 x 10-5 0.0752 0.011 PTCA 0.999 0.0441 0.000421 0.126 0.0633 DHI -melanin 0.999 13.0 0.205 0.157 0.889 DHICA-melanin 0.998 6.17 0.201 1.11 0.871 3AT 0.990 0.143 0.00627 0.0690 0.0961 AHP 0.988 0.208 0.00804 0.212 0.123 2-CysDOPA-melanin 0.995 1.21 0.0553 0.096 0.182 5-CysDOPA-melanin 0.987 0.203 0.00965 0.00684 0.0318 -- --- 41 VS. the amount of hair subjected to analysis. The limits of quantitation and (limits of detection) were as follows: DHI-melanin 10 Jlg (10 J.lg); DHICA-melanin 1 Jlg (0.50 Jlg); 2-CysDOPA-melanin 240 ng (120 ng); and 5-CysDOPA-melanin 170 ng (85 ng). (Occasionally, larger injection volumes were used to increase signal to noise up past limits of quantitation.) Degradation Product Cross-reactivity To assess melanin degradation product cross-reactivity, that is, production of degradation markers from melanins other than the primary producer of a specific degradation marker, eumelanins were subjected to pheomelanin analysis and vice versa. Results (shown in Table 2.3) demonstrate that PTCA is produced from DHI melanin and that PDCA and PTCA are produced from 5-CysDOPA and 2-CysDOPA pheomelanins. However, based on their precision (Table 2.3), cross-reactivity results are consistent and thus can be accounted for when calculating melanin composition. Analysis of Heteropolymeric Melanins To test the reliability of the chemical degradation assays in analyzing heteropolymeric melanins, a series of mixed composition eumelanins, mixed composition pheomelanins, and mixed eu-/pheomelanins were made and subjected to analysis. As an outside verification that the mixed eumelanins were of expected monomer composition, samples of each mixed eumelanin polymer were subjected to elemental analysis by cOlnbustion (after being dehydrated overnight at room temperature under a vacuum), Galbraith Laboratories, Inc., Knoxville, TN. As the number of DHI monomer units in the melanin is increased, one expects the percent carbon and percent nitrogen content of Table 2.3 Crossreactivity table-production of fnelanin nlarkers froln unexpected melanin sources. Values expressed as the ratio of nmol degradation product observed per nmol of expected degradation product 1:: standard deviation. n=4 for each dete nnination. PDCA PTCA AIIP 3AT DHI-melanin - 1.1 ± 0:'083 Not detected Not detected DHICA-melanin Not detected - . Not detected Not detected 5-CysDOPA-melanin 2.6 ± 0.052 5.8 ±O.ll - Not detected 2-CysDOPA-lnelanin 2.2 ± 0.11 7.6 ± 0.17 Not detected +:.. t...) 43 that melanin to increase. Shifts in elemental analysis results agree with the monomer composition results (as determined by chemical degradation analysis) as eumelanin monomer composition is shifted from DHI to DHICA: 75% DHICA /25% DHI melanin was 53.43% C and 7.72% N; 50% DHICA / 500/0 DEI melanin was 54.10% C and 7.740/0 N; 250/0 DHICA I 750/0 DHI melanin was 56.090/0 C and 7.83% N. Because the pheomelanin monomers are structural isomers of each other, one cannot tell via changes in elemental composition what the changes in pheomelanin polymer composition are. However, because the monomers are structural isomers, their relative incorporation into the polymer is expected to be approximately equal. Thus, due to lack of an available second verification procedure, an outside verification of pheomelanin subtype composition was not carried out. More eumelanin and less pheomelanin were found in the mixed eu-/pheomelanin polymer than was expected from the starting monomer composition. This was verified by elemental analysis results, which confirm that the sulfur content of the mixed eu-/pheomelanin heteropolymer mixture is less than 3.90/0. The theoretical dehydrated value is 5.4%. The presence of more eumelanin and less pheomelanin in this polymer mixture than was expected from starting monomer composition is thought to be due to the fact that pheomelanin is slightly soluble in the condi60ns under which the polymer was isolated once it had been formed. Overall, the fact that chemical degradation results agree with the actual synthetic melanin compositions (as verified by elemental analysis) (see Figure 2.4 for chemical degradation assay results) indicates that the assays can reliably assess the composition of heteropolymeric melanins. Figure 2.4A demonstrates that when eumelanins are synthesized from different monomer compositions, the alkaline hydrogen peroxide 44 Figure 2.4 Analysis of- A) Mixed eumelanin series, B) Mixed pheomelanin series, and C) Melanin made from 25% (w/w) of all four melanin subtypes. Results agree with elemental analyses and are expressed as calculated percent composition based on the mass of degradation product found, divided by the mass/mass % yield of that degradation product (expressed as a fraction) from a melanin made solely from the monomer that produces that degradation product, divided by the total mass of melanin degraded, and multiplying by 100 to obtain a percent value. n=4 for each mixed melanin. Error bars indicate standard deviation. A c: -0- -+-II f/) 0 c. E 0 0 0~ -c G) +II -cu ::J (.) -cu 0 100 II Calc. % DHICA Melanin 75 D Calc. % DHI T Melanin 50 25 o -+----1 - 7S% DHICA I 2So/0 DHI SOO/o DHICA I So% DHI 2So/0 DHICA I 7So/0 DHI ..J:::.. Vl B c: --..0.--. . tn 0 c. E 0 0 # 1J .C..D.. -co :::J -0co 0 100 75 - 50 25 o - II Calc. % 2- CysDOPA Melanin - D Calc. % 5- II CysDOPA Melanin 75% 2-CysDOPA 500/02-CysDOPA 250/02-CysDOPA I 250/0 5- I 50% 5- I 75% 5- CysDOPA CysDOPA CysDOPA ..J:::.. 0', c 100-T-~<~--~-'~"~-----=~-~~----'-----"---~--'-~~------__ ' __ "_'_'~"_n,._~ II Calc. % DHICA Melanin 75 0 Calc. % DHI Melanin II Calc. % 2-CysDOPA Melanin 50 II Calc. 5-CysDOPA Melanin 25 o -11----- 48 oxidation assay will provide accurate information regarding the composition of the heteropolymeric eumelanins. Likewise, Figure 2.4 B shows that when pheomelanins are synthesized from different monomer compositions, the reductive HI hydrolysis assay will provide accurate information regarding the composition of the heteropolymeric pheomelanins. And finally, Figure 2.4C shows that even when mixed eu-/pheomelanin copolymers are made, the eumelanin and pheomelanin assays can determine the relative composition of each melanin subtype. Melanin Composition of Human Hair Types The melanin composition of different colored hair from 44 donors was assessed (3 African-Americans with black hair, 1 American-Indian with black hair, 6 Asians with black hair, 6 Anglo-Americans with black hair, 2 Hispanics with black hair, 12 Anglo­Americans with brown hair, 8 Anglo-Americans with blond hair, and 6 Anglo-Americans with red hair) (Figure 2.5). Hair color determinations were made by visual inspection prior to melanin analysis. Results show that black-haired individuals (regardless of race) have the most DHI and DHICA melanin while red-haired Anglo-Americans have the most 5-CysDOPA and 2-CysDOPA melanin. Caucasians with black hair have less total melanin in their hair than other races with black hair examined in the study. To correlate hair melanin content to codeine (a basic drug, pKa 8.2, Figure 2.6) incorporation after a single oral dose, a scatter plot of total eumelanin vs. hair codeine content divided by plasma AVC was made (Figure 2 .. 7). (Experimental design and codeine content of these human hair samples was previously determined by Rollins et al. (126». A linear regression analysis (Microsoft Excel 2000) on these data produces an r2 coefficient of 0.73, meaning that 73% of the variation in the data can be accounted for by 49 Figure 2.5 Analysis of human hair for melanin subtypes: A) Eumelanin. *Significantly different from all black hair (p < 0.0001). tSignificantly different from African­American black hair (p < 0.01). tSignificantly different from Asian black hair (total and DHI melanin only) (p < 0.05). #Significantly different from brown hair (p < 0.05). B) Pheomelanin. <l>Significantly different from red hair' (total pheomelanin and both subtypes) (p < 0.01). See results section for the number of saInples from each group. Error bars indicate standard deviation. A Cl ~ 20 ::l ""-'" -c=: 15 -co ...-... (J) J.. E-cu 10 ::JJ: W 'fa- o ..... c: ::l o E « 5 -- o Black Hair (African­Arrerican) Black Hair (Asian) t:}: Black Hair (Hispanic) Hair Color (Race) D DHICA Melanin • DHI Melanin * Brovvn Hair VI o 51 -cc-:: .cc-:: -CO -CO ~ ~ <C.<C c..c.. 88 ~~ I I LnN o • 0 0 0 0 0 0 0 L.() 0 L.() or- ~ 52 Figure 2.6 Chemical structure of codeine 53 Figure 2.7 The relationship between codeine incorporation into human hair and eumelanin content. Codeine concentration is expressed in ng codeine / mg hair. AUC stands for the area under the plasma concentration vs. time pharmacokinetic curve. Dividing the codeine concentration in hair data by plasma AUC normalizes the codeine content of the hair for the amount of drug present in the blood. This procedure makes use of the assumption of a direct linear relationship between blood levels and hair concentrations of codeine, as demonstrated in rats by Gygi et al. (146). • • (VJIRa + IRa = U!uBI;;Jwn;;J) .lIBQ ~w / U!uBI;;Jwn;;J ~n -=: -=> 54 u p < ........ ~ .-.-,,=;--... Q.i -=- Q.i '"CS 0 U I...-' eunlelanin content alone. Pheomelanin content of human hair demonstrates no relationship with codeine content. Discussion 55 Several methods have been developed for quantitating the amount of total melanin, eumelanin, and pheomelanin in biological samples (130,132,141,144). The data presented in this study correlated amounts of melanin subtype-specific degradation markers to actual amounts of melanin subtypes in a sample. This was done by determining percent yield values for a given degradation marker from its parent melanin (homopolymer) SUbtype and using these values on melanin from hair samples to back­calculate how much of each melanin subtype was in the hair. Possible difficulties concerning the validity of the assays such as assay linearity, melanin SUbtype cross­reactivity, and assay reliability when assessing heteropolymeric melanins were addressed. It must be kept in mind, however, that DRI and DRICA are the major precursors to eumelanin, but eumelanin polymers do not necessarily consist solely of these monomers, even when these are the only monomers used in the synthesis of the polymer. This is due to the fact that hydrogen peroxide, produced in the vicinity of developing melanin polymers during the monomer oxidation process, can partially alter the structure of the forming polymer by oxidatively degrading some of the DRI and DRICA units into carboxylic acid substituted pyrrole units (95, 96). Nevertheless, eumelanins consisting mostly of DRICA units would be expected to have a greater carboxylic acid content than eumelanins consisting mostly of DRI units. (This hypothesis has been confirmed by Novellino et al. (95)). In addition, the procedure used to synthesize a melanin can affect the carboxyl content of the resulting melanin (95) which, in turn, may affect yields of PDCA and PTCA from the resulting polymers. Because of this concern, the melanins used in this study were synthesized via a method that best models the in vivo situation. That is, tyrosinase was used as the catalytic enzyme while the reaction was carried out under oxygen in the absence of catalase. 56 Using the assays and degradation product yield information as described above, it was determined that average black human hair contains approximately 99% eumelanin (600/0 DRI and 40% DHICA-derived eumelanin) and 1 % pheomelanin (80% 2-CysDOPA and 20% 5-CysDOPA-derived pheomelanin); brown and blond hair contain 95% eumelanin and 5% pheomelanin; and red hair contains 670/0 eumelanin and 33% pheomelanin (Figure 2.5). These data suggest that color determination for black, brown, or blond hair depends more on melanin quantity than eu-/pheomelanin composition, while red hair may arise through an alteration in the melanin synthesis pathway that leads to a greater relative production of pheome1anin. This knowledge must be considered when relating drug-eumelanin or drug-pheomelanin binding data to drug incorporation into hair. It has been found that yields of degradation markers are different for each sUbtype of melanin, but that yields are highly reproducible (Table 2.1) and thus can be used to assess the amount of melanin sUbtypes in a given sample. However, as shown in Table 2.3, some degradation markers are not always produced solely from their primary melanin sUbtype source, e.g., PTCA is not produced solely from DRICA melanin. Because of consistent yields, and the fact that DHICA-melanin is not cross-reactive, this phenomenon can be accounted for in calculating amounts of melanin sUbtypes in a sample. This phenomenon is disconcerting given the postulated structures (Figure 2.1) 57 for melanin sUbtype precursors, polymers of which have been shown to produce unexpected degradation products. Studies by Napolitano et al. (147, 148) have shown that DHI dimers linked at the 2- position can produce both PDCA and PTCA. It is quite interesting to note the 1: 1 PDCA:PTCA molar ratio produced from DHI melanin in light of the degradation pathway proposed by Napolitano et al. (147) whereby there is a 500/0 chance of producing either PDCA or PTCA from a 2-4 linked DIll dimer. Production of PDCA and PTCA from both 5-CysDOPA and 2-CysDOPA pheomelanins is more difficult to explain mechanistically. The production of PDCA and PTCA from pure pheotnelanins suggests the presence of indole units in the polymer. This could mean that monomers 1 and/or 2 shown in Figure 2.8 may exist as part of the pheomelanin polymer. Considering the nucleophilic nature of the phenylalanine side chain amino group and the fact that monomers 1 and 2 have not been ruled out as part of the make up of 5-CysDOP A and 2-CysDOP A pheomelanin, the presence of such monomers may explain the production of PDCA and PTCA from pheomelanins. According to Ito et al. (149), the existence of monomer 1 in the pheomelanin polymer should be verified by the production of cysteine in hydriodic acid hydrolysates of 5- CysDOPA and 2-CysDOPA pheomelanin. Using HPLC coupled to mass spectrometry (LCIMS) it was demonstrated that no cysteine is produced from 5-CysDOPA and 2- CysDOPA pheomelanin hydriodic acid hydrolysates (data not shown) thus likely ruling out the presence of monomer 1. Monomer 2, however, remains a possibility because it should produce 3AT or AHP upon hydriodic acid hydrolysis and PDCA and PTCA upon alkaline hydrogen peroxide degradation. d' Ischia et al. (150) identified monomer 2 (saturated at the 2- position) when cysteine was added to oxidized DOPA. They reported HO HO S N H r-COOH NH2 Monomer 1 (COOH) HOOC~ II S N HO Monomer 2 Figure 2.8 Chemical structures of monomers 1 and 2 58 (COOH) 59 that this material rapidly autooxidizes to give complex mixtures of yellowish-brown oligomeric materials. The existence of monomer 2 in pheomelanin~ however, remains to be proven. In conclusion~ the data presented in this paper provide the means to calculate the amount of each melanin sUbtype present in a hair sample. As demonstrated in Figure 2.7, a strong relationship exists between codeine incbrporation into human hair and eumelanin content. A similar relationship may exist for other basic drugs as well-perhaps providing a means to normalize drug concentrations in hair to eumelanin concentrations so as to eliminate any potential color bias. As testing for drugs in hair becomes a more commonplace procedure, the role of specific melanin sUbtypes may become more pronounced. Quantitation of melanin subtypes therefore may become critically important in interpreting quantitative drug testing results. CHAPTER 3 AMPHETAMINE AND N-ACETYLAMPHETAMINE INCORPORATION INTO HAIR: AN INVESTIGATION OF THE POTENTIAL ROLE OF DRUG BASICITY IN HAIR COLOR BIAS g Forensic testing for drugs of abuse in hair has become a popular alternative to traditional urinalysis. Hair color bias, however, presents a problem to fair hair testing. A number of basic drugs including cocaine (66,151), phencyclidine (69), codeine (65,152), stanozolol (68), and nicotine (40) have been found to incorporate to a greater extent into dark hair over light colored hair. To develop nonbiased testing procedures, the roles of drug chemistry and hair pigmentation in determining drug concentrations in hair must be understood. Hair pigment, or melanin, is a polyanionic indolequinone-based polymer that has the potential to interact with positively charged molecules. A number of researchers have suggested that drug-melanin binding through ionic and/or van der Waals interactions between drug molecules and melanin polymers maybe responsible for the preferential incorporation of some drugs into pigmented vs. nonpigmented hair (30,64,69,75, 153). Basic drugs are positively charged at physiologic pH. If such drugs are incorporated into gReproduced, with slight modification, from C.R. Borges, D.O. Wilkins, and D.E. Rollins. 1. Anal. Taxieol. 25: 221-227 (2001), (the Journal of AnalytiealToxieology) by permission of Preston Publications. A Division of Preston Industries, Inc. 61 hair they may bind to melanin and show a preference for incorporation into pigmented vs. non-pigmented hair. Neutral or acidic drugs may only be able to interact with melanin through weak van der Waals forces, thus they are less likely to show a preference for incorporation into pigmented vs. nonpigmented hair. In agreement with this idea, phenobarbital, an acidic drug, does not incorporate preferentially into pigmented vs. non­pigmented rat hair (65). In addition, other evidence indicates that carbamazepine (pKa -7) does not exhibit a hair color bias in humans (74). Although characterized as a basic drug, a relatively low pKa leaves most of the drug uncharged as it circulates throughout the body. Ionic interactions with melanin do not provide the only means for a drug to get into hair. In addition to N-AcAp (41) (data shown below), other nonionic compounds such as ~9-tetrahydrocannabinol (THC) (33-35, 154), and a large number of steroids such as testosterone (155-157), dehydroepiandosterone (158, 159), 17-p-estradiol (160), and nandrolone (161-163) have been found in hair indicating that melanin may not be the only drug binding site in hair (46, 49,63). Amphetamine and N-AcAp have been shown to incorporate into hair (3, 41). This paper describes the use of amphetamine (pKa 9.8) and its nonbasic analog N-AcAp (Figure 3.1) to investigate the role of drug basicity in the hair color bias that is sometimes observed in drug incorporation into hair. Materials and Methods Chemicals and Reagents d-Amphetamine sulfate and d-amphetamine sulfate-d3 were obtained from Sigma (St. Louis, MO). N-AcAp and N-AcAp-d3 were synthesized with acetic anhydride OINH2 I I ~ Amphetamine o~ OINH N -Acety lamphetamine Figure 3.1 Chemical structures of amphetamine and N-AcAp. 62 and acetic anhydride-d6 obtained from Sigma, according to the method of Halmekoski and Saarinen (164). Identities and purity of the crystalline products were verified by mass spectrometry and IH-NMR eH-NMR for dO N-AcAp only) (see Appendix). The melting point of d-N-acetylamphetamine and d-N-acetylamphetamine-d3 was 123-124 °C. N-butyl chloride, chloroform, and methanol (HPLC grade) were obtained from Burdick and Jackson (Muskegon, Mn. All other reagent grade chemicals were obtained from Mallinckrodt (St. Louis, MO). Stock Solutions and Preparation of Standard Curves Separate stock solutions of amphetamine sulfate and N-AcAp (5 mg/mL) in deionized water and methanol, respectively, were made. (Only the mass of amphetamine from the amphetamine sulfate was accounted for when calculating dilutions.) From stock solutions, working solutions of 0.1, 1, and 10 ng/IlL drug in water were made. Daily standard curves were obtained by analyzing hair samples (20 mg) fortified with amphetamine or N-AcAp at 0,0.1,0.2,0.3,0.5, 1,2,3,5,10,25, and 50 ng/mg hair. Quality control (QC) samples (0.3, 3, and 25 ng/mg hair) were prepared daily from working solutions of stock sources different from those used to prepare standard curves. Animal Protocols Male LE rats (--150 g) were obtained from Harlan Sprague-Dawley (Indianapolis, IN) and maintained in an environmentally controlled room with a 12-h light-dark cycle and free access to food and water. Animals were housed individually in hanging wire cages to prevent contamination from urine, saliva, bedding, or other rats. Drug free rat hair was obtained by shaving nondosed LE rats. Prior to dosing, the entire dorsal area of 64 each rat was shaved. d-Amphetamine sulfate or N-AcAp dissolved in normal saline or corn oil, respectively, was administered by intraperitoneal injection once daily at a dose of 10 mg/kg (n=8 for each drug) for 5 days. Regrown hair was shaved and separated into white and black samples 14 days after dosing was started. Hair samples were again collected 28 days after the initial drug administration. Hair was stored at -20°C before digestion and extraction. Each hair sample was carefully analyzed under a large magnifying glass and any skin flakes present were removed. Sample Digestion and Extraction Hair samples were mulched with scissors and then weighed (-20 mg) into separate 16 x 100 silanized glass test tubes. After the addition of 2 ng/mg d3 internal standard, samples were completely solubilized in 2 mL 1 M NaOH overnight at 37°C. To each sample was added 200 JlL concentrated ammonium hydroxide, followed by 4 mL n-butyl chloride: chloroform (4: 1), Samples were then mixed for 30 min on a test tube rocker, followed by centrifugation for 10 min at 1800 rpm in an IEe International (Needham Heights, MA) swing bucket centrifuge. The organic layer was transferred to a separate test tube and dried under a gentle stream of air at room temperature. Recoveries of amphetamine and N-AcAp were approximately 800/0, thus no significant loss of analyte occurred during evaporation. The dried extracts were reconstituted in 100 J.LL LC mobile phase for injection. LCIMSIMS Analvsis HPLC coupled to tandem mass spectrometry (LC/MS/MS) analysis was selected as the analytical technique due to its excellent selectivity and sensitivity as well as its 65 ability to analyze polar compounds without derivitization. Analysis was performed in positive ion mode on a ThermoQuest TSQ/SSQ 7000 mass spectrometer (Finnigan, San Jose, CA) equipped with an electrospray ionization interface coupled to a Hewlett Packard (Palo Alto, CA) 1100 Series HPLC. Spray voltage was set at 4.5 kV, heated capillary at 250°C, collision gas at 3.5 mTorr, and collision offset at -25 V. HPLC separation was performed on a Phenomenex (Torrance, CA) Luna 30 x 2.00 mm, 5 J..Lm, C18(2) column under the following co~ditions: Amphetamine was eluted isocratically with 80% 0.1 % formic acid I 20% methanol and monitored at its 136 mlz to 91 mlz transition. Amphetamine-d3 was monitored at its 139 mlz to 92 mlz transition. N-AcAp was eluted isocratically with 40% 0.1 % fonnic acid I 60% methanol and monitored at its 178 mlz to 91 mlz transition. N -AcAp-d3 was monitored at its 181 mlz to 91 rnJz transition. Statistical Analysis Comparisons between groups were made with a two-tailed Student's t-test assuming homogeneity of variances. Results significant at p < 0.05 are reported with relevant approximate p values. Results Assay Validation To validate the analytical method for amphetamine and N-AcAp, intra- and interassay accuracy and intra- and interassay precision were evaluated (Table 3.1). Data for amphetamine and N-AcAp recovery are also included in Table 3.1. Accuracy, precision, and recovery experiments were done with stock solutions separate from those Table 3.1 Method validation for amphetamine and N-AcAp. Inter assay precision was detennined with three runs three to five samples each. See Results section for explanations of how percentages were calculated. Amphetamine n Target 0/0 N-AcAp n Target 0/0 Concentration Concentration Intra Assay Accuracy 0/0 n=5 0.3 ng/mg 6.0 Intra Assay Accuracy 0/0 n=5 0.3 ng/mg 3.5 n=5 3 ng/mg 2.6 n=5 3 ng/mg 3.2 n=5 25 ng/mg 0.9 n=5 25 ng/mg 6.7 Inter Assay Accuracy 0/0 n=15 0.3 ng/mg 6.6 Inter Assay Accuracy % n=11 0.3 ng/mg 7.1 n=15 3 ng/mg 1.5 n=14 3 ng/mg 6.9 n=15 25 ng/mg 2.8 n=11 25 ng/mg 8.3 Intra Assay Precision % CV n=5 0.3 ng/mg 7.6 Intra Assay Precision % CV n=5 0.3 ng/rTl9 1.8 n=5 3 ng/mg 0.9 n=5 3 ng/mg 0.7 n=5 25 ng/mg 1.5 n=5 25ng/mg 11.8 Inter Assay Precision % CV n=15 0.3 ng/mg 10.0 Inter Assay Precision % CV n=11 0.3 ng/mg 7.0 I n=15 3 ng/mg 0.8 n=14 3 ng/mg 1.5 I n=15 25 ng/mg 5.1 n=11 25 ng/mg 5.2 0/0 Recovery n=5 0.3 ng/mg 85.6 . % Recovery n=5 0.3 ng/mg 74.0 n=5 3 ng/mg 79.7 n=5 3 ng/mg 81.0 n=5 2§Qg/rl}g 80.7 n=5 25 ng/mg 80.0 -~- 0\ 0\ 67 used to prepare standard curves. Accuracy was calculated by dividing the absolute value of the difference between the target concentration and the observed concentration by the target concentration then multiplying by 100 to obtain a percentage. Data are thus read as being accurate to within a certain percentage of the target concentration. Precision was assessed as the percent coefficient of variation from the mean concentration. Recovery was determined by spiking samples with internal standard after extraction and dividing their mean observed concentrati'on by the mean observed concentration of another group of samples to which internal standard was added as described above. Amphetamine and N-AcAp were quantitated by obtaining peak area ratios (dO/d3) and comparing these to a standard curve created in the same manner. Accuracy remained within 8.3% and precision remained within 11.8% at the concentrations studied here. Standard curves were linear from 0 to 50 ng / mg hair and l values were typicaliy 0.99 or higher. The limit of quantitation (LOQ), set at a minimum signal to noise ratio of 5 and quantitation accuracy of 20%, was 0.1 ng drug/mghair. Recovery of both amphetamine and N-AcAp from the extraction procedure was approximately 80%. Amphetamine and amphetamine-d3 were monitored with 136 m1z to 91 m1z and 139 m1z to 92 m1z transitions, respectively, and N-AcAp and N-AcAp-d3 were monitored with 178 m1z to 91 m1z and 181 m1z to 91 m1z transitions, respectively. Representative selected reaction-monitoring (SRM) chromatograms for amphetamine and N-AcAp found in rat hair are shown in Figure 3.2. Chromatographic runs were 3 min in length. Hair collected prior to dosing (day 0) contained no detectable drugs. Day 28 hair for both amphetamine and N-AcAp was at or below the LOQ. Analysis of hair collected 14 days after dosing (Figure 3.3) revealed that amphetamine"was incorporated into black hair at a 68 Figure 3.2 Representative SRM chromatograms for A) amphetamine found in rat hair and B) N-AcAp found in rat hair A 1 0.0 0.5 SRM: 136 m/z to 91 mJz SRM: 139 m/z to 92 mJz 1.0 1.5 Time (min) 69 B 0.0 0.5 SRM: 178 nllz to 91 m/z SRM: 181 In/z to 91 nl/z 1.0 1.5 Time (min) 70 2.0 2.5 3.0 Figure 3.3 Concentrations of A) amphetamine found in white and black rat hair and B) N-AcAp found in white and black rat hair. Error bars represent standard deviation. n=8 for each drug. *Significant at p < 0.00 1. 71 (j)C:Of'....C'OL()~Cf)C\J-r-O J!ell 6w/6u [aU!We~alldw,,] 72 73 or- Q). ex:>. f'... CD. L!). ~. (1). C\I. or-. 0 000000000 J!ell 6w/6u [9U!WeJ9I1dweIAI9:>V-N1 mean concentration of 6.44 ± 1.31 (SD) ng amphetamine / mg hair, an amount significantly different (p < 0.001) from that found in white hair (2.04 ± 0.58 ng amphetamine/mg hair) (n=8). In contrast, no difference in N-AcAp content was found between black hair (0.87 ± 0.08 ng N-AcAp/mg hair) and white hair (0.83 ± 0.15 ng N­AcAp/ mg hair) from rats dosed with N-AcAp (n=8). Discussion 74 Nakahara et al. (41,43,67, 75) have suggested that drug basicity, which, in theory, leads to drug-melanin binding, is an important characteristic governing the incorporation of a drug into hair. Additional studies have indicated that although drugs do not have to be basic (e.g., phenobarbital) to be highly incorporated into hair (65), drug basicity may playa strong role in the preferential incorporation of a given drug into pigmented vs. nonpigmented hair (65, 66,68, 69, 127, 153). To directly evaluate the role of drug basicity in potential hair color bias, this study employed amphetamine and N­AcAp (a basic drug and its nonbasic analog) to test the hypothesis that a basic drug will show a hair color bias, but that a nonbasic drug will not show a hair color bias. LE rats were chosen for this study because the presence of black and white hair on the same animal helps eliminate black hair vs. white hair drug concentration differences due to additional factors such as species differences and interanimal variability. (Also, by simpJe visual inspection, no differences were noted between black hair and white hair growth rates on the same anima1.) The results of this study clearly demonstrate that drug basicity can lead to a hair color bias, but a lack of basicity appears to eliminate hair color bias. Melanin likely plays the dominant role in the demonstrated hair color bias, but we did not attempt to normalize hair concentrations for melanin content due to the fact that while there certainly is melanin in black LE rat hair, there are negligible amounts of melanin in white LE rat hair (as detennined by the HPLC method of Borges et al. (165), chromatograms not shown). Mathematically, nonnalizing the data in this study to hair melanin content would produce meaningless data. futerestingly, the fact that drugs were incorporated into white LE rat hair demonstrates that melanin is not the only binding location for drugs in hair. The difference in basicity between amphetamine and N-AcAp represents the most striking contrast between these two drugs, but there is also a difference in lipophilicity, a feature that may be partly responsible for the differences in hair incorporation noted in these studies. However, the greater lipophilicity of N-AcAp is probably not responsible for its lack of a hair color bias or its lesser incorporation into hair regardless of hair color. This is because lipophilicity neither increases melanin binding (see Chapter 4) or overall incorporation into hair, as might be expected according to classical hair incorporation theories. Evidence for hair color bias also exists for cocaine (66), phencyclidine (69, 128), nicotine (40), cotinine (40), stanozolol (68), and codeine (64, 65) but not for the acidic drug phenobarbital, which shows no hair pigmentation bias whatsoever (65). Melanin consists of polyanionic indolequinonone-based polymers (96) that have been partially degraded into carboxylic acid substituted pyrrole units (95). In theory, the hair color bias demonstrated for basic drugs may be caused by the ionic binding of positively charged, basic drug molecules to negatively charged melanin pigment polymers. In vitro drug­melanin binding studies with amphetamine and N-AcAp were also conducted to further test this theory (see Chapter 4). 76 Some epidemiological evidence, (72, 73, 166) has suggested that there is no hair color bias for amphetamines and cocaine. These authors suggest that differences in hair drug concentrations between races are due to differences in drug preferences among the various societal groups rather than due to a hair color bias (defined as the situation where more drug is deposited into darker hair vs. lighter hair, given equal dosing). One primary conclusion is that, " ... there is no significant relationship between [hair] color category and likelihood to test positive for cocaine. (166)" Unfortunately, information on exactly which individuals from different races actually ingested the drugs in question is not available in these studies. Hair analysis simply says whether or not there is drug in hair above a certain concentration. As such, there may not be an effective hair color bias. Such epidemiological evidence is derived from a dichotomous hair test outcome, the results of which depend on an established cutoff concentration above whi.ch a test result is considered positive. It may be the case that the cutoff concentration is set appropriately so that all (or nearly all) of the individuals taking the drugs in question are found positive regardless of hair color. That is, for (nearly) all users, sufficient drug is incorporated into their hair to produce a positive test outcome regardless of hair color. There may have been no or too few users exposed to a small enough dose for light-haired individuals to test negative and dark-haired individuals to be shown positive for drug use. Importantly, while there may be more drug in dark colored hair, there may still be enough drug in light colored hair to produce a positive result. A recent study by Rollins et ai. (126) helps to elucidate this matter. Their study demonstrated clear differences between light and dark colored hair concentrations of ofloxacin and codeine for a group of human subjects given the same low therapeutic dose of ofloxacin or codeine. Thus, they 77 demonstrated hair color biases for both ofloxacin and codeine testing in hair at therapeutic doses. As clearly shown by their results, however, for a given dose, an unfair testing situation only exists if the cutoff concentration is set too high. That is, if the cutoff concentration is set above the mean drug concentration in light hair but below the mean drug concentration in dark hair. While it is necessary to set high enough cutoff concentrations to clearly distinguish from background noise, setting cutoff concentrations as close to background levels as possible will help to make any possible unfair hair color effects as minimal as possible (Le., there will be fewer individuals in the dose range where a negative test will be produced for light colored hair and a positive test produced for dark colored hair). The likelihood of eliminating an unfair hair-testing situation, however, would have to be determined for each individual drug. In comparing the overall incorporation of amphetamine vs. N-AcAp in hair, Nakahara et at. have demonstrated using dark agouti rats that even though plasma concentrations of N-AcAp are much higher than those of amphetamin.e (41) (given equal doses), amphetamine is incorporated into hair to a far greater extent than is N-AcAp (41). Their findings regarding amphetamine and N-AcAp in hair were substantiated by this study in LE rats (in both black and white hair). Thus while drug-melanin binding certainly plays a strong role in drug incorporation into hair and determining hair color bias, additional processes such as hair cell drug transport may be important determinants of the amount of a given drug incorporated into hair-pigmentation differences aside. As hair testing for drugs of abuse becomes more commonplace, the relationships between drug and hair cell biochemistry must become better understood to allow for fair and accurate interpretation of drug testing results. CHAPTER 4 COCAINE, BENZOYLECGONINE, AMPHETAMINE, AND N-ACETYLAMPHETAMINE BINDING TO MELANIN SUBTYPES Melanin is an indolequinone based polymer that is polyanionic in nature and thought to be capable of binding drugs through both'ionic and possibly van der Waals interactions (51, 53). In vivo, melanin is formed linked to proteins in organelles known as melanosomes (pigment granules), which are produced in melanocytes. Melanocytes ultimately transfer their melanosomes to keratinocytes where the pigment can be evenly displayed on the body surface or in hair (96, 167). The binding of a wide variety of drugs to melanin both in vitro and in vivo is well documented (51-53, 55, 79, 168), Because melanin is found throughout the body in areas as diverse as the inner ears, eyes, skin, hair, and brain, the pharmacological and toxicological consequences of drug-melanin binding range from reduction in antibacterial activity (169), induction of drug toxicity (87), and protection from drug toxicity (85, 86), to implications for the origin of Parkinson's disease (77, 78, 80, 88, 89, 170-172). Our laboratory is primarily interested in drug-melanin binding from the perspective of how it affects drug incorporation into hair. A number of studies have shown that hair pigmentation plays a major role in the concentration of drug incorporated into hair (63, 68, 69, 127). These data suggest that a hair color bias may exist when 79 testing for drugs of abuse in hair. That is, because more drug is incorporated into the hair of a dark-haired individual, their hair is more likely to be found positive for a drug that binds melanin than is the hair of an individual with light colored hair. While there is some debate as to whether or not melanin can produce an effective "hair color bias" in actual drug testing (72, 73), the fact that drugs can bind to melanin remains uncontested. To date, the only drugs definitively shown to incorporate to a greater extent in dark hair over light hair have been basic drugs-i.e., drugs that have a positive charge at physiological pH. Negatively charged and neutral drugs such as phenobarbital and N­AcAp specifically investigated for a hair color bias have been shown to lack this phenomenon (42, 65). Such findings suggest that it is the ionic binding between positively charged drug molecules and negatively charged melanin polymers that leads to a greater incorporation of a drug into pigmented vs. nonpigmented hair. In fact, this suspicion is supported by mechanistic studies on the nature of drug-melanin interactions (51,53). Melanin, which is ultimately derived fron1 the amino acid tyrosine (Figures 2.lA and 2.lB), can be divided into two types: the black eumelanins and the reddish-brown pheomelanins. These types can be further divided into SUbtypes: Eumelanins are made up of primarily DHI and DHICA monomers, (Figure 2.lA) and pheomelanins consist of primarily 5-CysDOPA and 2-CysDOPA-derived monomer units (Figure 2.lB). DHI and DHICA are the major precursors of eumelanin, but the polymer does not necessarily consist solely of these monomers. This is due to the fact that hydrogen peroxide, produced in the vicinity of developing melanin polyn1ers, can partially alter the forming polymer by oxidatively degrading some of the DHI and DHICA units into carboxylic 80 acid substituted pyrrole units (95, 96). Nevertheless, eumelanins consisting mostly of DHICA units would be expected to have a greater carboxylic acid content than eumelanins consisting mostly of DHI units. (This hypothesis has been confirmed by Novellino et al. (95)). Because of the significant chemical differences between DHI and DHICA eumelanins and the fact that pheomelanins are substantially chemically different from eumelanins, the binding properties'of different melanin subtypes may differ for a given drug. The purpose of the work presented here was to assess the roles of individual melanin sUbtypes in binding amphetamine, N-AcAp, cocaine, and BE (Figures 3.1 and 4.1). These drugs were chosen because cocaine (pKa 8.5) exhibits a hair color bias (63, 66) while its zwitterionic chemical congener, BE, has not been shown to exhibit this bias. Similarly, while amphetamine exhibits a hair color bias (42), its nonbasic analog N-AcAp has been shown not to exhibit a hair color bias (42). In addition to the drug-melanin binding studies, attempts were made to begin to elucidate the chemical functional groups on melanin responsible for drug binding. Several model molecules containing one melanin functional group each were analyzed via tandem mass spectrometry for adduct formation with amphetamine and cocaine. The information gleaned from these studies helps to elucidate the important chemical properties of both drugs and melanins that are important for drug-melanin binding and thus for producing a potential hair color bias. 81 Cocaine Benzoy lecgonine Figure 4.1 Chemical structures of cocaine and BE 82 Materials and Methods Materials DRI and DRICA were synthesized as described by Wakamatsu and Ito (142). DHI-me1anin and DRICA-me1anin were prepared according to the method of Ito et al. (144). 5-CysDOPA and 2-CysDOPA were made via the method of Ito et al. (143). CysDOPA-me1anin and 2-CysDOPA-me1anin were prepared as described by Ito and Fujita (141). Mixed eu-/pheomelanin copolymers from DRI and 5-CysDOPA or 2- CysDOPA were also made by the method of Ito and Fujita (141) omitting the acetone wash step. The initial reaction mixture contained 40% DRI and 60% CysDOPA (by mass). Melanin analysis by the method of Borges et. al. (165) revealed that the copolymers consisted of 80% DHI-me1anin and 20% 5-CysDOPA-melanin or 2Q% 2- CysDOPA-melanin (by mass), 3R-1-(-)-Cocaine was obtained from NEN Life Science Products, Inc. (Boston, MA). 3R-1-( -)-Benzoylecgonine was obtained from Moravek Biochemicals (Brea, CA). Tyrosinase, 1-(-)-cocaine hydrochloride, benzoyl ecgonine tetrahydrate, d-amphetamine sulfate, d-amphetamine-d3 sulfate, and acetic anhydride-d6 were obtained from Sigma (St. Louis, MO). N-AcAp and N-AcAp-d3 were synthesized according to the method of Ra1mekoski and Saarinen (164), using acetic anhydride-d6 for N-AcAp-d3. Aminopropy1si1ica (APS) solid phase sorbent was purchased from International Sorbent Technology (Mid Glamorgan, UK). Cytoscint ES liquid scintillation fluid was obtained from ICN Biomedica1s, Inc. (Costa Mesa, CA). Buffer salts were obtained from Mallinckrodt (Paris, KY). The structure and purity of all synthetic compounds (except melanins) were confirmed by NMR and mass spectrometry (See Appendix). Matrix assisted laser desorption ionization mass spectrometry (MALDI- MS) on pure pheomelanin was performed on a PerSeptive Bioscience Voyager-Elite mass spectrometer equipped with a time-of-flight mass analyzer in positive ionization mode using as trans-4-hydroxy-3-methoxycinnamic acid as the sample matrix, by Dr. Vajira Nanayakkara. All other chemicals used are readily available and were of the highest purity available. Melanin Preparation 83 Five different melanins were used in these experiments. They included: DHI­melanin, DHICA-melanin, 80% DHI/ 20% 5-CysDOPA-melanin (DHIJ5-CysDOPA­melanin), 800/0 DHI / 20% 2-CysDOPA-melanin (DHIJ2-CysDOPA-melanin) and 100% 5-CysDOPA-melanin covalently linked to an aminopropylsilica (APS) solid phase sorbent, hereafter referred to as APS-pheomelanin. 5-CysDOP A was covalently attached to the APS sorbent using the same technique that Ibrahim and Aubry (173) used to link tyrosine-deri ved melanin to APS. Melanins were prepared for binding experiments by repeated homogenizing and washing in 0.066 M potassium phosphate buffer, pH 7.4 (for cocaine and BE binding experiments) or 0.15 M ammonium acetate buffer, pH 7.4 (for amphetamine and N-AcAp binding experiments) until a clear supernatant was obtained. Buffer concentrations were chosen to match the ionic strength of a normal saline solution. Ammonium acetate buffer was used for amphetamine and N-AcAp samples because a volatile buffer was required for LCIMSIMS analysis. Binding Experiments For each drug with each melanin, a time course (5, 15, 30, 60, and 90 min) was run to determine the amount of time required for binding equilibrium to be reached. Time course samples were run at a concentration of 500 pM for cocaine, 2 nM for BE, and 30 nM for amphetamine and N-AcAp. Binding equilibrium was always established by 45 or 60 min. To obtain the binding parameters Bmax and Ka for each drug-melanin combination, a range of drug concentrations was incubated with each melanin. For cocaine and BE this range included 12 points between 300 pM and 75 ~M. For amphetamine and N-AcAp this range included 12 points between 1 nM and 300 /lM. 84 Binding experiments were conducted in silariized 16 x 100 mm glass test tubes by adding the specified amount of drug (dissolved in buffer) to a buffer solution into which had been added 100 flg melanin (from a 1 mg/ml melanin suspension in buffer). The preparatory procedure was the same for all melanins except APS-pheomelanin for which it was necessary to add 3 mg APS-pheomelanin to equal 0.1 mg actual pheomelanin. Incubation was then performed at 37 °C in a vigorously shaking water bath, keeping melanin gran,ules in suspension (325 rpm) for the specified time. For each sample, an analogous sample was incubated in the absence of melanin to serve as a control for any possible background binding. (F9r APS-pheomelanin samples, blank, or unreacted APS, was used as the background.) Total incubation volumes were 1 mL. At the end of the incubation time, samples were immediately transferred to 1.7 rnL microcentrifuge tubes and spun at 14,000 rpm for 90 min at 4 °C. Immediately after centrifugation, 0.6 rnL of supernatant was removed and mixed with either 14 inL scintillation fluid (cocaine and BE) or an aliquot of internal standard (amphetamine-d3 or N-AcAp-d3) at approximately one-third the amount of drug used in the incubation. Cocaine and BE samples were then mixed, left to settle overnight, and counted in a Packard 1900CA Tri-carb liquid scintillation counter. Amphetamine and N-AcAp samples were evaporated in a Savant (Holbrook, NY) Speed-Vac SPD121P concentrator at < 35°C. Samples were then reconstituted in 100 JlL LC mobile phase for analysis. LC/MS/MS Analysis 85 Quantitative analysis. LC/MS/MS analysis was performed in positive ion mode on a ThermoQuest (Finnigan, San Jose, CA) TSQ/SSQ mass spectrometer equipped with an electrospray ionization interface coupled to a Hewlett Packard (Palo Alto, CA) 1100 Series HPLC. Spray voltage was set at 4.5 kV, heated capillary at 250°C, argon collision gas at 3.0 mToIT, and collision offset at -25 V. HPLC separation was performed on a Phenomenex (Torrance, CA) Luna 30 x 2.00 mm, 5 Jlm, C 18(2) column under the following conditions: Amphetamine was eluted isocratically with 85% 0.1 % formic acid I 15% methanol and monitored at its 136 mlz to 91 mlz transition. Amphetamine-d3 was monitored at its 139 mlz to 92 mlz transition. N-AcAp was eluted isocratically with 45% 0.1 % formic acid I 55% methanol and monitored at its 178 mlz to 91 mlz transition. N-AcAp-d3 was monitored at its 181 mlz to 91 mlz transition. Qualitative analysis. Qualitative tandem mass spectrometry experiments were performed to determine which melanin functional groups are most important in drug binding. Amphetamine and cocaine were assessed for noncovalent adduct formation with the functional groups on melanin as modeled by oxidized catechol, reduced catechol, deprotonated phthalic acid, and 3-methylindole. Tandem mass spectrometry analyses were carried out on the same instrument as above with the following alterations: Spray voltage was set to 3.5 kV, argon collision gas at 2.0 mToIT, and collision offset varied between -12 and -18 V. Positive ions were monitored for analysis of solutions of drug molecules mixed with oxidized catechol, reduced catechol, and 3-methylindole. 86 Negative ions were monitored when looking for adducts in mixtures of drugs with phthalic acid. Experiments were carried out by mixing the drug and model compound of interest at 1 00 ~ and 200 ~M final concentrations in 0.1 % formic acid (unless otherwise specified), respectively, then infusing the mixture into the electrospray source at 1 0 ~min with a supplementary HPLC flow of methanol (unless otherwise specified) at 0.05 rnUmin. Oxidized catechol was made fresh daily by oxidizing a 0.02 M solution of catechol in water with a few milligrams of Ag20. The resulting yellow-brown mixture was then centrifuged at 14,000 rpm in a microcentrifuge at 4 °C for 5 min. The supernatant was then desalted by solid phase extraction as follows: A Waters Sep-Pak C18 solid phase extraction column was conditioned with 3 mL of methanol followed by 3 rnL of water. One milliliter of approximately 0.02 M oxidized catechol was applied to the column, which was then washed with 9 mL water. Oxidized catechol was eluted with 2 rnL of methanol to obtain a solution of oxidized catechol at approximately 0.01 M. Phthalic acid was kept deprotonated by keeping it in aqueous solution brought to pH 9 with ammonium hydroxide. The supplementary HPLC flow for experiments with phthalic acid was 50% water 150% methanol at 0.05 mL/min. If an adduct was detected, daughter ion mass spectra were obtained for the parent adduct ion and neutral loss spectra scanning for any expected neutral losses were obtained to confirm the presence of both the drug molecule and the functional group model as part of the parent ion. To confirm adducts, experiments were run with deuterated drugs to confirm expected adduct mass shifts. 87 Data Analysis For cocaine and BE the amount of free drug in the form of free fraction of drug was obtained by dividing the counts per min (CPM) for a melanin-containing sample by the CPM for its analogous nonmelanin-containing san1ple. For amphetamine and N­AcAp this was done by dividing the dO:d3 chromatographic peak area ratio for a melanin­containing sample by the dO:d3 chromatographic peak area ratio for a nonmelanin­containing sample. The bound fraction of drug was then taken as the difference between 1 and the free fraction of drug. The free concentration of drug (in M) was then determined by multiplying the free fraction by the concentration of drug in the sample. The bound concentration of drug (in moles drug/mg melanin) was determined by multiplying the bound fraction by the amount of drug in the sample and dividing this by the amount of melanin in the sample (0.1 mg). To obtain the binding parameters Bmax (binding capacity in moles drug / mg melanin) and Ka (binding affinity in M-1 ), data from a range of drug concentrations were plotted as [Bound] vs. [Free] and analyzed via nonlinear least squares regression analysis using the equation describing drug adsorption to a solid (a Type I Langmuir Isotherm) (174), That is, where [Bd] is the concentration of bound drug, Bmax is the maximum binding capacity of the melanin (in moles drug / mg melanin), Ka is the drug-melanin association constant (in M-1 ), and [D] is the concentration of free drug (in M). 88 To determine the number of different types of binding sites (n) Scatchard plots were made. Scatchard plots indicated that for drugs that bound to melanin there were generally two types of binding sites, high affinity/low capacity and low affinitylhigh capacity binding sites. Nonlinear least squares regression analysis provided the low affinitylhigh capacity binding parameters and the x-intercept and (negative) slope of a Scatchard plot of the low concentration data (first four points) provided the binding parameters for the high affinity/low capacity binding site. Two-tailed Student's t-tests assuming homogeneity of variances were used to compare binding parameters for a drug between melanins. Results To investigate the binding of cocaine, BE, amphetamine, and N-AcAp to a variety of different melanin subtypes, drugs were incubated with melanins first at various time points to determine the time required to reach binding equilibrium, then at various concentrations to determine Bmax and Ka (as described above). Representative selected reaction-monitoring (SRM) chromatograms from analysis of amphetamine and N-AcAp are shown in Figure 4.2. Table 4.1 presents a summary of the binding parameters determined for each drug with each melanin. Bmax represents binding capacity in moles of drug per mg melanin and Ka represents binding affinity in M-I . Experiments were repeated three times for drugs that bound to melanin and twice for drugs that did not bind to melanin. Out of the four drugs tested for binding to DHI, DHICA, DHII5-CysDOPA Figure 4.2 Representative SRM chromatograms from LCIMSIMS analysis of (AJ amphetamine and (BJ N-AcAp 89 A 0.0 0.5 SRM: 136 mlz to 91 mlz SRM: 139 mlz to 92 mlz 1.0 1.5 nme (min) 90 2.0 2.5 3.0 B t) t) a 1 "'0 § .0 < t) > ~ '3 0.0 0.5 SRM: 178 rnJz to 91 rnJz SRM: 181 m/zto91 rnJz 1.0 1.5 Time (min) 2.0 91 2.5 3.0 Table 4.1: Drug melanin-subtype binding parameters.a Drug binding to DHI-melanin High Affinity / Low Capacity Low Affinity / High Capacity log Ka -log Blnax log Ka -log Brnax N = Number of Binding Sites Cocaine 8.2±0.19 9.5 ± 0.17 4.9 ± 0.039 7.0 ± 0.078 2 Amphetamine 6.6 ± 0.40 9.3 ± 0.35 3.9 ± 0.34 7.0 ± 0.25 2 Drug binding to DHICA-melanin High Affinity / Low Capacity Low Affinity / High Capacity log Ka -log Bmax log Ka -log Bnlax N = Number of Binding Sites .. Cocaine 7.8 ± 0.14 9.6 ±"0.16 4.5 ± 0.29 7.1 ± 0.082 2 Amphetamine 7.2 ± 1.2 9.9 ± 1.2 NBb 1 a Ka in units of M-1 . Bmax in units of mol drug/mg melanin. n=3 for all points except n=2 for NBs and n=4 for Ap-DHI-melanin and Ap-80% DIU / 20% 2-CbsDOP A-melanin binding studies. NB indicates no binding observed. ! \0 N Drug binding to 80% DHI/ 200/0 5-CvsDOPA ~ '" High Affinity / Low Capacity Low Affinity / High Capacity logKa -log Bmax logKa -log Bmax N = Nurnber of BindtnQ" Sites Cocaine 8.0 ± 0.31 9.5 ± 0.29 c4.3 ± 0.23 6.8 ± 0.21 2 Amphetamine COMpd 4.1 ± 0.32 7.1 ± 0.25 2 Drug binding to 80% DB! /20% 2-CysDOPA-rnelanin High Affinity / Low Capacity Low Affinity / High Capacity log Ka -log Bmax log Ka -log Brnax N = Number of Binding Sites Cocaine 8.2 ± 0.048 9.6 ± 0.061 c4.4 ± 0.17 6.9 ± 0.16 2 Amphetamine COMpd 3.7 ± 0.32 6.9 ±0.29 2 a Ka in units of M-1 • Bmax in units of mol drug/mg melanin. n=3 for all points except n=2 for NBs and n=4 for Ap-DHI-melanin and Ap-80% DRI I 20% 2-CbsDOPA-melanin binding studies. NB indicates no binding observed. c Indicates a significant difference (p < 0.05) from DHI-melanin. d COMP inidcates competition from buffer cations. I \0 w 94 DHIJ2-CysDOPA, and 5-CysDOPA melanins, only cocaine and amphetamine were found to bind to melanins. BE and N-AcAp did not-bind to any of the melanins in this study. In addition, cocaine and amphetamine were found not to bind to the 5-CysDOPA­melanin (when APS was used as a measure of background binding, data not shown). As a positive control for the APS-linked 5-CysDOPA pheomelanin binding studies, DOPA-derived melanin was linked to the APS solid phase and found to bind cocaine with similar characteristics as "free" L-DOPA melanin. For most of the (tested) drug-melanin binding interactions, there appeared to be two different binding sites; a high affinityllow capacity binding site and a low affinity/high capacity binding site. The concave nature of Scatchard plots derived from binding data demonstrated this to be the case. A representative Scatchard plot for cocaine binding to DHI-melanin is provided in Figure 4.3. Note that a group of data points appears to run along the y-axis and another group appears to run more parallel with the x-axis. The former section represents the high affinityllow capacity binding site (Bmax and Ka values were determined by the x-intercept and (negative) slope of the section, respectively), while the latter section represents the low affinity/high capacity binding site. Because linearized (double reciprocal) equations and their associated plots tend to skew the relative error of the smallest plot values by disproportionately relying upon them as the major determinants of plot slope and intercept values, Bmax and Ka values for the low affinity/high capacity binding sites (a region defined by higher concentration data points) were determined through nonlinear least squares regression analysis as discussed above and demonstrated in Figure 4.4. Figure 4.3 Scatchard plot for cocaine binding to DHI-melanin. Two linear sections are present, indicating two different types of binding sites. Error bars indicate standard deviation. N = 3. 95 t-+-1 I-+-f CD LO ~ ('t) C\J .,.... 0 0 0 0 . 0 0 0 0 0 0 0 0 1I\I.6w/IOW [aaJ.:I]/[pu n08] I 0 "o"" I W C\J .,.... a:> 0 I ill 0 CD 0 0 + W 0 0 96 en E .-......... 0 ,.E..... -c c ::J 0 .r.n... . m E ::::;;; o E ...... "0 £: ::s o .a..t.. m o ..J 10-10 10.12 o 97 Cocaine Binding to DHI-Melanin .................... ".':' .............. '"' . ........... ':" ..................... . ····················t···· .. ···················t········ ················~ .. ·"················'····f········"···············t············· .. · .. ····· . ··t······ .... ····· .. · ...... l l 1 1 1 1 .................. ..l. ..................... ..1. ..................... ..1. ...................... ..1 ..................... ..1. ....................... L.. I I ! I ! ! ···················r·····················r······················I························r···················r .. ········· .. r·················· ···················r·····················r·····················T······················r .. ··················r······················r················· iii ! 1 ! 1 10.5 [Free] M Figure 4.4 Hyperbolic binding data plot for cocaine binding to DHI-melanin. Such plots were used to perform nonlinear least squares regression analysis and determine Bmax and Ka values for the low affinity/high capacity binding sites. Due to the wide range of drug concentrations used, data are plotted on a semilog scale so that most of the data points do not appear to be clustered at the origin. Error bars indicate standard deviation. n = 3. 98 As shown in Table 4.1, there were relatively few significant differences in the binding parameters for a given drug to the different melanin sUbtypes. The major difference was that none of the drugs that bound to eumelanins and mixed eu­Ipheomelanins bound to pure (APS linked) pheomelanin. A direct comparison of Ka values between drugs is not legitimate due to the fact that different buffer systems were used. Ka data are not directly comparable because different buffer cations may compete differently for melanin binding sites and thus alter the Ka values that may have been observed if the binding studies had been conducted in identical buffer systems. Comparison of binding parameters between cocaine .and amphetamine analogs, however, was not a goal of this study-as this was done previously for cocaine and amphetamine (55). Except in cases where binding data could not be assessed for amphetamine, Bmax values for cocaine and amphetamine were similar (Table 4.1). This is in agreement with Shimada and co-worker's data (55). Competition for melanin binding from buffer cations, as indicated by a convex Scatchard plot (51) (Figure 4.5), was found to occur for amphetamine binding to mixed eu-/pheomelanins, but not to pure eumelanins. When binding experiments were conducted in the absence of buffer, the apparent competition from buffer cations was eliminated and a concave Scatchard plot was demonstrated. The only other significant difference between melanins was the lower binding affinity of cocaine to the low affinity/high capacity binding site of the mixed eu-/pheomelanins compared to the analogous site on pure eumelanins (Student's t-test, p < 0.05). Noncovalent amphetamine and cocaine adducts with oxidized catechol, reduced catechol, deprotonated phthalic acid, and 3-methylindole were searched for by tandem Figure 4.5 Scatchard plot for amphetamine binding to a mixed eu-Ipheomelanin (80% DHI I 20% 5-CysDOPA-melanin). Due to the wide range of drug concentrations used, data are plotted on a semilog scale so that lower x-axis values can be distinguished from each other. The convex shape indicates competition for binding from buffer cations. Error bars indicate standard deviation. n = 3 99 100 i I I i ,.- 0 0 0 t !-- 0 + 0 0 . 0 I+t ~ f C) ttl ::E::: 0 I' E r-"I ! 0 "'C ,.- t: ~ I ::J W 0 t ,.- .C...D... . ! .. ~ I .41 ... I I "'llllJI"'" I - - I I I I (V) -- ,.- I W Ct) LO C\J LO ,.- L.() 0 ,.- 0 C\J 0 ,.- 0 0 0 0 0 0 0 0 0 0 . 0 0 0 0 0 0 0 WJf.6w/IOW [aaJ.:I]/[punos] 101 mass spectrometry as a means of assessing which chemical functional groups on melanin are most important for drug binding. Mass spectra shown in Figure 4.6 show the peaks corresponding to the only drug-functional group adduct observed; that of amphetamine with a covalent dimer of oxidized catechol (350 mlz) (Figure 4.7). (Other peaks in the single stage spectra below 130 mlz and at 332 mlz and 360 mlz are artifacts from the solvents employed.) The proposed structure shows a 3-3' linkage between the monomers as this seems the most likely candidate given that monon1ers of oxidized catechol do not form adducts with amphetamine; i.e., the oxygen atoms of the two monomers need to be in close proximity to each other to effect adduct formation. Other linkages such as 3-4' and 4-4', as well as neutral semiquinone radicals may be involved in adduct formation, but it is impossible to determine from the data obtained. Mass spectra shown in Figure 4.6 (D)-(F) demonstrate the mass shift of the adduct observed when amphetamine-d3 was employed in the study instead of amphetamine. MS/MS daughter ion and neutral loss spectra in Figure 4.6 (B)-(C) and (E)-(F) also show that amphetamine is part of the parent ion and that the covalent dimer of oxidized catechol (neutral 214 amu) is lost whether the experiment was performed with amphetamine or amphetamine-d3. Amphetamine in solution with reduced catechol produced no peaks that would correspond to amphetamine adduct formation with reduced catechol or a reduced catechol dimer. Discussion Binding of cocaine and amphetamine to synthetic L-DOPA derived melanin has been demonstrated previously (55, 56). Data from Shimada et at. (55) obtained from drug binding studies to L-DOPA melanin in 0.1 M potassium phosphate buffer, pH 7.4, indicate log Ka and -log Bmax values for (±)-cocaine of 5.9 and 8.6, respectively, and for 102 Figure 4.6 Mass spectra providing evidence for the amphetamine-oxidized catechol dimer adduct. (A) through (C) represent spectra obtained from a 0.1 % formic acid solution containing 100 flM amphetamine and -200 flM oxidized catechol. (D) through (F) represent spectra obtained from a 0.1 % formic acid solution containing 100 JlM amphetamine-d3 and -200 flM oxidized catechol. Single stage mass spectrum (A), MSIMS of 350 rnJz (B), neutral loss scan for 214 amu (C), single stage mass spectrum (D), MSIMS of 353 rnJz (E), and neutral loss scan for 214 amu (F). A 135.9 150 200 250 nl/Z Single Stage Scan + 350.1 350 103 B 1~ .l 136.1 J as...; '1 [ ()Ii-or "'I J ~ 50 200 250 m/z MSIMS: Daughter Ion Scan of350 m/z + 350.0 104 c 350.1 + 220 240 21JO . 300 320 340 380 380 . 400 mlz MSI1vfS: Neutral Loss of214 Scan I ' • 420 440 460 105 D 138.9 [ ()Y~-DJ 250 mJZ Single Stage Scan 353.0 106 E 139.1 ! ... 50 ! I i 100 I i i ~ 1 i { i 150 i I 250 mJZ 300 MSIMS: Daughter Ion Scan of353 mJz 353.1 !, 350 I ' 400 107 • I 450 F 100, 280 280 353.1 + , , 380 . , 340 360 mlz MS/M:S: Neutral Loss of214 Scan 108 109 Figure 4.7 Proposed amphetamine-oxidized catechol dimer adduct as detected by tandem mass spectrometry. Energy minimized spacefill diagram showing the interaction of positively charged amphetamine with the oxidized catechol dimer. 110 (+)-amphetamine of 5.4 and 8.7, respectively. The data of Shimada et al. were obtained through the use of double reciprocal plots, which only reveal binding data for the high affinity/low capacity binding site. A direct comparison between their Ka data and ours is not legitimate due to the fact that different buffer systems were used, but it can be seen from Table 4.1 that their Bmax values are reasonably close to the values obtained in this study. The greater ionic strength of their buffer system relative to ours, however, suggests that they should observe lower apparent Ka values-which they did. Because cocaine and amphetamine have previously been found to bind to melanin, it was not surprising to find that these drugs bind to most of the melanin SUbtypes in this study. The chemical congeners of cocaine and amphetamine, BE and N­AcAp, respectively, however, were found not to bind to any type of melanin at all. This information provides clues as to what chemical features are important for drug-melanin binding. The structural difference between cocaine and BE is a methyl ester vs. a carboxylic acid. Based on suspected mechanisms of drug-melanin binding (51, 53), the . electronic properties of a methyl ester are unlikely to be important for drug-melanin binding. Rather, the positive charge on the nitrogen of cocaine (pKa 8.5) is most likely to be responsible for the drug's binding to melanin. For BE, the introduction of a negative charge in the same molecule could render the positively charged nitrogen much less available for ionic binding with the negatively charged melanin polymer. In a similar scenario, the acetyl group of N-AcAp makes the amphetamine molecule nonbasic and thus not charged at physiologic pH. Again, it can be seen that the effective removal of a positive charge renders the drug molecule unable to bind to melanin. 111 Assuming the importance of a positively charged drug molecule for binding to melanin it is expected that the greater number of negatively charged atoms in the melanin polymer, the more positively charged drug molecules would bind to it (i.e., Brnax should increase). As stated in the introduction, a comparison of DHI and DHICA monomers suggests that DHICA melanin should have a greatercarboxylic acid content (and thus stronger anionic nature) than DHI melanin. This idea was confirmed by Novellino et al. (95) when it was demonstrated that DHICA melanin produces twice as much CO2 as DHI melanin upon acid catalyzed thermal decarboxylation of the melanin pigments. Thus it was expected that DHICA melanin would have a larger capacity (Brnax) for binding a given positively charged drug than DHI melanin. This was found not to be the case. Reasons for this are not certain, but there are a several possibilities. First, it may be possible that the carboxylic acid group is not responsible for cocaine or amphetamine binding, which seems unlikely given the mechanistic evidence for ionic binding interactions (51, 53). Second, a simple visual inspection (as well as visible wavelength optical absorbance-data not shown) of the homogenized DHI and DHICA melanin suspensions revealed that for a given mass/volume concentration of melanin in aqueous suspension, DHI melanin has more surface area (is finer and less grainy) than DHICA melanin. This may render a relatively significant portion of the carboxylic acid groups in DHICA melanin unable to interact with solvent molecules and thus unable to bind to drug present in solution. Drug molecules seem to bind best if they are positively charged, regardless of whether or not they have a hydrophobic region such as a phenyl group (compare amphetamine and N-AcAp binding). In addition, a simple nickel (II) cation can display 112 the same complex (multi-site) binding characteristics as a structurally more complex drug molecule (51). This study and those of others (51, 53) suggest that there must be two separate sites on melanin that are capable of binding a positively charged ion or molecule. It has been previously suggested that the ortho-semiquinone free radical centers in melanin can bind doubly and triply charged cations forming chelate complexes (83). This could account for one of the binding sites for N?+, but due to the fact that nitrogen atoms within organic molecules cannot act as the guest of a ohelation complex, it cannot account for the binding of drug molecules. Tandem mass spectral analysis of a solution of amphetamine mixed with oxidized catechol (Figure 4.6) strongly suggests that amphetamine can interact with the orthoquinone groups on oxidized indolequinone units of melanin. Interestingly, amphetamine was not found to complex with oxidized catechol monomers. This suggests that melanin secondary structure may play an important role in drug-melanin binding. This hypothesis is supported by the fact that cocaine did not form any adducts with oxidized catechol as amphetamine did, even though cocaine also demonstrates biphasic (two-site) binding to melanin. Molecular dynamics energy minimization experiments using the ChemSite ChemDemo program with a bath temperature of 300 K, an integration time step of 1 fs, total time of lOps, 100 equilibration steps, and non-bonded interaction cutoff distance of loA show that the hydrogens attached to the primary amine of amphetamine intercalate between the oxygens of the dimerized oxidized catechol molecule and likely form three separate hydrogen bonds as depicted in Figure 4.7. The fact that cocaine did not form adducts with oxidized catechol is consistent with this model, because the positively charged tertiary amine of cocaine could not intimately 113 interact with the oxidized catechol dimer in the same manner as amphetamine due to steric hindrance. Based on the data obtained in these studies in conjunction with the investigations of others, it seems that the high affinity/low capacity binding site for basic drugs with melanin arises from an interaction of the positively charged drug molecule with appropriately shaped, electron dense, secondary structures of the melanin polymer. Evidence that this may be the high affinity/low capacity binding site also comes from Stepien and Wilczok (53) who demonstrated that adding a substantial am9unt of ethanol (20-500/0) to a binding experiment selectively eliminates the high af