Studies of the pharmacokinetics, cellular actions and motor stimulatant effects of caffeine in Horses


Material Information

Studies of the pharmacokinetics, cellular actions and motor stimulatant effects of caffeine in Horses
Physical Description:
Chou, Chi-Chung, 1966-
Publication Date:

Record Information

Rights Management:
All applicable rights reserved by the source institution and holding location.
Resource Identifier:
aleph - 27827454
oclc - 54703324
System ID:

Full Text







This work is dedicated to my father Ting-Li Chou, and the Chou families of this
generation, with expectations of more to come.


First and foremost I would like to express my deepest gratitude to my major

advisor Dr. Thomas Vickroy for his exceptional guidance and support. My great gratitude

also goes to my co-chairman Dr. Alistair Webb who always stood by me and helped me

keep perspective. Without them, my graduate education would not have continued.

I extend much gratitude to my advisory committee members, Dr. Murray Brown,

my friend and mentor, for his help and valuable suggestions as I moved through the

research; Dr. Ronald Gronwall and Dr. William Millard for their intellectual support,

constructive criticism and encouragement. Their knowledge and expertise were

indispensable for the successful accomplishment of my study.

I am very grateful to Dr. Pat Colahan, for his generosity in supporting me

financially and sharing his lab facilities and resources throughout my graduate studies. I

am also grateful to Mr. Brett Rice and Mrs. Marty Johnson for their companionship and

assistance during my experiments.

My sincere thanks also go to Dr. Chao-Lin Chen, who introduced me to the

University of Florida for my advanced education.

Finally, this dissertation would not have been possible without my wife, Karen.

Her sacrifice and support in every way possible throughout this long, long process has

been the most significant factor behind the scenes.



ACKNOW LEDGM ENTS ................................................................................................. iii

ABSTRACT...................................................................................................................... vii


Introduction to Caffeine.................................................................................................. I
Overview ..................................................................................................................... 1
Pharmacokinetics of Caffeine..................................................................................... 3
Pharmacological Effects of Caffeine.......................................................................... 6
Cellular M echanisms of Caffeine Action................................................................... 7
Antagonism of adencdsine receptors........................................................................ 8
Inhibition ofphosphodiesterase............................................................................ 11
M obilization of intracellular calcium .................................................................... 11
Interactions with dopaminergic and benzodiazepine receptors............................ 12
Caffeine Use and Athletic Performance................................................................... 14
Introduction to M icrodialysis Technique...................................................................... 16
Principals of M icrodialysis....................................................................................... 18
M icrodialysis Systems .............................................................................................. 19
M icrodialysis Probe and Recovery.......................................................................... 21
Quantitation of M icrodialysis Data........................................................................... 26
Validation of probe recovery ................................................................................ 26
Analytical considerations...................................................................................... 31
Limitations of M icrodialysis.................................................................................... 32
Application of M icrodialysis .................................................................................... 35
M icrodialysis in Free-M oving Large Animals ......................................................... 36
Conclusion ................................................................................................................ 37

SEMI-AUTOMATED IN VIVO MICRODIALYSIS .................................................. 39
Introduction................................................................................................................... 39
M materials and M ethods........................................................................................... 41
Experimental Subjects .............................................................................................. 41
M icrodialysis Apparatus and Probe Implantation..................................................... 41
Drug Administration and Sample Collection............................................................ 43

Drug Analyses........................................................................................................... 44
Pharmacokinetic Analyses........................................................................................ 45
Data Analyses and Statistics..................................................................................... 46
Results.......................................................................................................................... 47
In Vitro Calibration of M icrodialysis Probes............................................................ 47
Comparison of CA Concentrations in Venous Blood by Venipuncture and In Vivo
M icrodialysis............................................................................................................. 48
Comparison of CA Concentrations in Venous Blood and Splenius Muscle............ 50
CA Metabolites in Mixed Venous Blood and Splenius Muscle............................... 51
Discussion .................................................................................................................... 52

STIMULATION IN EQUINE FOREBRAIN TISSUE ................................................ 67
Introduction................................................................................................................... 67
M material and M ethods ................................................................................................... 70
M materials ................................................................................................................... 70
M embrane Preparation.............................................................................................. 71
Binding of [3H]DPCPX to M embranes..................................................................... 71
Binding of [3H]ZM 241385 to M embranes ............................................................... 72
Binding of [35S]GTPyS to M embranes..................................................................... 73
Data Analysis............................................................................................................ 74
Results........................................................................................................................... 75
[3H]DPCPX and [ 3H]ZM241385 Binding Studies in Equine Forebrain................. 75
Competition Binding of M X at Adenosine Receptors.............................................. 76
Agonist and Antagonist Action on [35S]GTPyS Binding.......................................... 76
Discussion..................................................................................................................... 77

TREATED HORSES .................................................................................................... 95
Introduction................................................................................................................... 95
M materials and M ethods................................................................................................. 97
Experimental Subjects ............................................................................................. 97
Jugular vein and CSF M icrodialysis......................................................................... 97
Gentamicin Study........................................................................................ ....... 99
Drug Administration and Sample Collection............................................................ 99
Drug Analyses...................................................................................................... 100
Evaluation of Spontaneous M otor Activity............................................................ 100
Pharmacokinetic Analyses...................................................................................... 102
Data Analyses and Statistics................................................................................... 103
Results ......................................................................................................................... 103
Caffeine and CA Metabolite Concentrations in Venous Blood and CSF measured
by In Vivo M icrodialysis........................................................................................ 103
Penetration of Gentamicin Across Blood Brain Barrier......................................... 105


Distribution of CA and Metabolites into Central Nervous System........................ 105
Stimulant Effect of CA on Motor Activity............................................................. 106
Correlation Between CA Levels and Motor Activity............................................. 107
D discussion ................................................................................................................... 107

5 DISCUSSION AND CONCLUSION.......................................................................... 124
Conclusion and Future Directions .............................................................................. 139

APPENDIX HORSES AND THEIR USES IN EXPERIMENTS ................................. 143

R E FE R EN C E ...................................................................................................................144

BIOGRAPHICAL SKETCH ........................................................................................... 173

Abstract of Dissertation Presented to the Graduate School
of the University of Florida in Partial Fulfillment of the
Requirements for the Degree of Doctor of Philosophy



Chi-Chung Chou

August, 2001

Chairman: Thomas W. Vickroy, Ph.D.
Major Department: Physiological Sciences

In view of the wide use of caffeine (CA) as a central nervous system stimulant

and its potential to improve athletic performance in animals, the motor stimulant effects

of CA and its central mechanisms of action were studied in the horse. An in vivo

microdialysis procedure for free-moving horses was developed and used in conjunction

with measurements of locomotor activity to monitor the concentrations of CA and CA

metabolites in venous blood, splenius muscle and cerebral spinal fluid (CSF). After

intravenous administration of 3 mg/kg CA, pharmacokinetic and statistical analyses

indicated that microdialysis and venipuncture provide equivalent measures of drug levels

in the blood. Free CA levels in muscle were significantly higher and eliminated more

rapidly than CA in venous blood. Caffeine and theophylline (TP) crossed the blood brain

barrier at different efficiencies while theobromine (TB) was not detected in CSF. These

results suggested the suitable use of microdialysis for simultaneous measurement of drug

and metabolite concentrations and underscored the importance of direct measurements of

drug levels within target tissue. Caffeine dose-dependently (0.5 to 3 mg/kg) increased

motor activity for 2 to 4 hours; the motor stimulation correlated well with free CA levels

in venous blood but not to CA or TP levels in CSF.

While high-affinity and high-density adenosine Al receptor binding sites were

identified in both equine cortex and striatum, adenosine A2a binding sites were found

predominantly in striatum. Caffeine, TP and paraxanthine (XN) bound non-selectively at

adenosine Al and A2a receptors with micromolar affinities and exhibited rank order

potency of TP>XN>CA at both receptor subtypes. The exchange of [35S]-guanosine-5'-

(y-thio)-triphosphate binding studies illustrated the antagonistic nature of CA and

metabolites at adenosine receptors.

The studies have demonstrated that after a therapeutic dose of CA, the locomotor

activities significantly increased in the horse and levels of CA and TP reached in CSF

were sufficient to partially block equine adenosine Al and A2a receptors in vitro. It is

likely that adenosine receptor antagonism is involved in the motor stimulant effects of

CA in the horse. The developed microdialysis model is expected to be of value for

pharmacokinetic, pharmacodynamic and drug residue studies in large animals.


Introduction to Caffeine


In humans, caffeine (CA; 1,3,7 trimethylxanthine) is one of the most widely used

psychoactive drugs. Caffeine is a natural constituent in the leaves, seeds or fruits of more

than 63 plant species worldwide. People have consumed caffeinated beverages since

ancient times. References as early as 2,700 B.C. indicated that the Chinese Emperor Shen

Nung sipped hot brewed tea. For most people in the western world, coffee is the principle

source for daily dietary CA. The origins of coffee consumption by humans date back to

575 A. D. when, in Africa, beans were used as money and food. Use of coffee as a

beverage dates back to 850 A. D., when an Arab goat-herder named Kalde observed

night-long friskiness and frolicking in goats that had eaten the berries of coffee arabica

(New Encyclopedia Britannica, 1974). He then made a beverage from the plant and

proclaimed his discovery to the world. Until the 17th century, the world's limited supply

of coffee was produced entirely in Yemen. From then on, the popularity of coffee

increased tremendously. This beverage was used as an aid in enduring long religious

ceremonies by the Muslims (Syed, 1972), and was very popular during the Renaissance,

supposedly providing inspiration and insight to the creators of many masterpieces.

Venetian traders introduced coffee to Europe. The first coffeehouse in London was

opened in 1652. Coffee drinking had been introduced into France nine years earlier, and

by 1690 there were 250 registered coffeehouses. Nevertheless, it was not until after the

Boston Tea party that extensive coffee drinking became prevalent in the American

Colonies (Johnson et al., 1975). Today, in the US alone, over two billion pounds of

coffee are consumed annually resulting in the annual consumption of well over 100,000

tons of CA (Dews, 1984). Approximately 80% of the US adult population consumes CA.

Mean CA intake in children 6 tol 1 months old is 4.2 milligrams (mg) per day; in children

6 to 17 years old, it increases to 43 mg per day (Graham, 1978). In adults, the average

American drinks 1.52 cups of CA-containing drinks per day, while the total CA intake is

approximately 206 mg per day (Dews, 1974). Coffee is not the only available source of

CA. Tea, which comes from the leaves of Thea sinensis, contains both theophylline (TP)

and CA. Cocoa and chocolate which come from fermenting seeds of the Theobroma

cocoa plant, contain both theobromine (TB) and CA. Among CA-containing beverages

that are most widely used by the general population are cola beverages, which contain

extracts of kola nuts. In addition to caffeinated drinks, approximately 2,000 non-

prescription drugs contain CA, including weight control aids, headache and pain

remedies, cold products, diuretics and at least 29 "stay awake" aids (Dews, 1974; Burg,

1975). Finally, at least one thousand prescription and non-prescription medications

contain CA, which is used as an ingredient in some pain relievers and is often combined

with salicylates or propoxyphene.

Caffeine consumption today is a fact of life all over the world. Although we enjoy

the desirable stimulatory effects of CA, the less desirable effects cannot be ignored.

Caffeinism or chronic CA consumption, which is generally recognized after daily intakes

of over 250 mg, can produce nervousness, irritability, tremulousness, occasional muscle

twitching, palpitations, flushing, hyperventilation, arrhythmias, tachypnea, tachycardia,

diuresis and gastrointestinal disturbances (Greyden, 1974). These symptoms occur

readily following ingestion of sufficient amounts of over-the-counter (OTC) medications

or CA-containing beverages.

Though CA consumption is not commonly considered an addictive drug, it should

be noted that CA is a drug with documented pharmacological actions. Numerous

researchers have identified relationships between CA and a number of cancers (Spiller

and Bruce, 1998) but none have established a cause-effect relationship. The consumption

of CA-containing products in moderation appears to be safe although limited ingestion is

recommended for those with diseases or conditions that may be aggravated by CA.

Pharmacokinetics of Caffeine

The pharmacological response to any agent depends not only on the potency and

efficacy of the drug, but also on its pharmacokinetic properties which include the rates of

absorption, distribution to tissues, metabolism to active and inactive metabolites, and

ultimate excretion from the body. In humans, CA absorption from the gastrointestinal

tract is rapid and reaches 99% bioavailbility usually within 45 minutes following

ingestion (Marks and Kelly, 1973; Bonati et al., 1982; Blanchard and Sawers, 1983;

Amaud, 1993). Caffeine absorption is also complete in rats, mice, and Chinese hamsters

following oral administration (Amrnaud, 1976, 1985; Aramaki et al., 1991), which gives

rise to comparable pharmacokinetic profiles irrespective of the route of administration

(Amrnaud, 1993; Aramaki et al., 1991). The source or form of CA does not seem to

influence drug absorption insofar as similar blood levels of CA are achieved following

equivalent doses of CA in beverages, capsules or chocolate (Mumford et al., 1996).

However, incomplete absorption of CA has been reported from coffee (Morgan et al.,


The hydrophobic properties of CA facilitate rapid passage through biological

membranes and near complete equilibration with total body water. Several studies have

shown that CA readily crosses the blood-brain barrier (BBB) in adult and fetal animals

(Lachance et al., 1983; Tanaka et al., 1984) and there is no long-term accumulation of

CA in brain tissue. The blood-to-plasma ratio of CA is close to unity (McCall et al.,

1982), which is consistent with limited plasma protein binding (< 15%) and free passage

into blood cells. Saliva concentrations of CA, which are considered to be a reliable index

of plasma CA levels, reach 65% to 85% of plasma concentrations (Cook et al., 1976;

Khanna et al., 1980). Peak plasma concentrations of CA are reached between 15 and 120

minutes after oral ingestion in humans. A dose of 1 mg/kg, roughly the equivalent of one

cup of coffee in an adult human, yields plasma levels of 1 to 2 p.g/mL or 5 to 10 jIM

(Arnaud and Welsch, 1982; Daly, 1993). Similar results have been obtained in horses

where 2.5 mg/kg gave rise to peak plasma CA concentrations of 2.5 to 4 gg/mL (12 to 20

ttM) (Aramaki et al., 1991, 1995, 1996; Peck et al., 1997).

Caffeine is metabolized mainly in the liver of human and animal by enzymes of

the cytochrome P450 system (1A2, 2E1) to produce more than 25 dimethyl and

monomethyl derivatives ofxanthine and uric acid (Somani and Gupta, 1988; Rodopoulos

and Nouman, 1996). Since cytochrome P450 (1A2) is expressed in brain and kidney,

these organs may also participate in CA metabolism (Goasduffet aL, 1996; Agundez et

al., 1998). Caffeine metabolism begins with removal of one of the three-methyl groups

on the xanthine ring. Mono-demethylation reactions of CA at positions 1, 3, or 7 produce

three major metabolites: which include TB (3, 7-dimethylxanthine), paraxanthine (XN;

1, 7-dimethylxanthine), and TP (1, 3-dimethylxanthine), respectively. The metabolic

routes for CA are remarkably species dependent. In humans, the predominant metabolite

is XN (84%) with markedly lower amounts of TB (10%) and TP (4%) (Kalow and Tang,

1993). In contrast to humans, TP is the major metabolite in horses (Aramaki et al., 1991)

while TB and TP prevail in rodents (Daly, 1993).

Caffeine is eliminated primarily in the form of metabolites in the urine. In

humans, less than 4% of an oral administration of CA is excreted as the parent drug in

urine (Somani and Gupta, 1988; Sinclair and Geiger, 2000). Clearance of CA follows

first order kinetics in most species, although saturation kinetics occurs in rats at doses of

40 mg/kg and above (Daly, 1993). At dose rates below 10 mg/kg, plasma CA half-lives

range from 0.7 to 1.2 hours in rats and mice, 3 to 5 hours in monkeys (Bonati et al., 1985)

and 2.5 to 4 hours in humans (Amrnaud, 1987). In spite of the relative short half-live (t )

in these and other species (Bonati et al., 1985; Kalow, 1985), horse exhibits a t V that is

greater than 13 hours. (Aramaki et al., 1991; Peck et al., 1997). In view of the fact that

the difference in CA elimination depends primarily on biotransformation, CA

pharmacokinetics in horses is likely to be related to a different rate or pathways for

metabolism. Common factors that affect P450 activity in general and CA metabolism in

particular in human include obesity, age, exercise, smoking, pregnancy, and the presence

of other drugs (Somani and Gupta, 1988; Swanson et al., 1997). The CA metabolites TP

and TB exhibit half-lives that are similar to CA in several species (Leo et al., 1986)

although these agents distribute preferentially into lung and respiratory tract tissues

(Aramaki, 1991) and displayed a reduced capacity to penetrate the BBB relative to CA

(Stahle etal., 1991; Daly, 1993). Nevertheless, in view of the marked pharmacological

activities of TP and TB, the actions of these metabolites must be taken into account when

considering the overall actions of CA. The pharmacokinetic difference between CA and

these two metabolites may explain their different uses in veterinary medicine where TP

and TB were used more often as anti-asthmatics rather than as stimulants.

Pharmacological Effects of Caffeine

Caffeine has pronounced effects in the central nervous system (CNS) as well as

cardiovascular, respiratory, renal, and muscular-skeletal tissues. Caffeine is considered

to be a CNS stimulant because of its ability to enhance many CNS functions. Caffeine

increases energy metabolism in the brain but decreases cerebral blood flow (Nehlig et al.,

1992, 1994). In humans CA administration increases locomotor activity (Hirsh, 1974;

Thithapandha et. al., 1972; Kaplan et al., 1989, 1990, 1991; Nehlig et al., 1992),

vigilance and attention (Land and Phillips-Bute, 1998); and delays onset of fatigue

(Graham et al., 1994). Other CNS effects of CA that are less readily quantified include

mood changes (Land and Phillips-Bute, 1998; Quinlan et al., 2000; Smit and Rogers,

2000), increased aggression (Sakata et al., 1973), anxiety (Sawyer et al., 1982; Fredholm

etal., 1992, 1999) and disruption of normal sleep patterns (Yanik et al., 1987; Snel,

1993; Fredholm et al., 1999). The purported ability of CA to enhance learning and

memory is not well documented (Nehlig et al., 1992; Fredholm et al., 1999). Finally, CA

produces a variety of neuroendocrinal effects including reduced levels of thyroid

stimulating hormone (TSH), and thyroxine and with elevations in growth hormone (GH),

cortisol, corticotrophin (ACTH) and P3-endorphin (Henry and Stephens, 1980; Spindel et

al., 1980; Arnold etal., 1982; Schlosberg, 1984).

In the cardiovascular system, CA increases coronary blood flow and elevates

systemic levels of norepinephrine, epinephrine and renin (Benowitz, 1990, 1995).

Caffeine also exerts a direct positive inotropic effect on cardiac muscle (Carrill and

Benitez, 2000; George, 2000) that is accompanied by an increase in transient blood

pressure, heart rate, and cardiac output (Wilcox, 1990; Nurminen et al., 1999). In the

respiratory system, both CA and TP exert anti-asthmatic effects (Rail, 1990; Daly, 1993;

Bara and Barley, 2000) primarily by relaxation of bronchial and tracheal smooth muscle

and through inhibition of the release of inflammatory mediators (Daly, 1993). In addition

to bronchodilation, CA and TP also stimulate respiration likely through central pathways

mediated by dopamine (Wessberg et al., 1985) and serotonin (Schmidt et al., 1995;

Spiller, 1998). Under conditions of strenuous exercise, CA may increase breathing

capacities through increasing V02 max (Wiles et al., 1992), minute ventilation (D'Urzo

et al., 1990) and CO2 sensitivity. In the renal system, both CA and TP increase renal

blood flow and glomerular filtration, stimulate renin release from the kidneys and

promote diuresis (Tofovic et al., 1996; Spill, 1998). In the muscular-skeletal system, CA

enhances lipolysis (Hetzler et al., 1990) leading to higher availability of free fatty acid as

muscle substrate during work, thus allowing glucose conservation (glycogen sparing

effect) (Nehlig and Debry, 1994; Vergauwen et al., 1997). Caffeine is also able to

increase skeletal muscle contractility (Ogawa 1994; Herrmann-Frank et al., 1999;

Tokutomi et al., 2001). Both the glycogen sparing effect and increased muscle

contractility appear to be beneficial to endurance muscular work.

Cellular Mechanisms of Caffeine Action

Three main hypotheses have been formulated to account for the actions of CA at a

biochemical or molecular level. These are, in the order in which they were reported,

mobilization of intracellular calcium, inhibition of phosphodiesterase (PDE) activity, and

antagonism of adenosine receptors. Based on studies in multiple animal species, it

appears that most investigations now favor the hypothesis that adenosine receptor

antagonism plays the most significant role. Inhibition of PDE and mobilization of

calcium, which occur at higher CA levels, may also contribute to several effects produced

by CA. For example, the respiratory and cardiac stimulant effects of CA appear to be

linked more closely to its PDE-inhibiting actions than to adenosine receptor antagonism

(Howell et al., 1990; 1993a). The effects of CA on CNS-mediated behaviors are

complex and do not appear to be fully attributable to blockade of adenosine receptors.

The possible involvement of other receptors and neurotransmitters within the CNS is

feasible though not as well characterized. For example, CA binds to benzodiazepine

receptors (Marangos et aL., 1979; Bouleuger et al., 1982) and may affect the regulation of

cellular functions that are regulated by the CNS neurotransmitter y-aminobutyric acid

(GABA). Caffeine also affects norepinephrine, dopamine, serotonin, acetylcholine, and

glutamate systems in brain (Daly, 1993) and, as a consequence of these actions; these

could contribute to the spectrum of pharmacological effects produced by CA.

Antagonism of adenosine receptors

Currently four distinct adenosine receptors, Al, A2a, A2b, and A3, have been

cloned and characterized in several species including human, rat, mouse, dog, and cattle

(Fredholm et al., 1994). Based on the binding affinities of these receptors for the

endogenous agonist adenosine, it appears that Al and A2a adenosine receptors would be

tonically activated at the normal extracellular adenosine concentrations that are present in

human and rat brain. Alternatively, A2b and A3 receptors exhibit much lower binding

affinities for adenosine and, therefore, are unlikely to be tonically activated (Fredholm,

1994; Fredholm et al., 1999). In view of these observations, Al and A2a receptors may

be considered to be most strongly affected by adenosine receptor antagonists such as CA

and related methylxanthines (MX) (Johansson et al., 1997). Both Al and A2a receptors

are guanine nucleotide proteins (G-proteins) in most cells. The Al subtype is coupled to a

pertussis-toxin-sensitive G-protein that mediates receptor-stimulated inhibition of adenyl

cyclase. In addition, Al receptors have been linked to other cellular actions including

inhibition of voltage-sensitive Ca++ channels (Hawke et al., 2000), opening of K+

channels (Olah and Stiles, 1995) and stimulation of phospholipase C and D activities

(Roberts-Thomson et al., 2000). The A2a receptors are coupled to a pertussis toxin

insensitive Gs-protein therefore the activation of this receptor subtype causes the

activation of adenyl cyclase as well as less characterized effects on voltage-sensitive Ca'+

channels. The Al receptors are widely distributed in almost all brain areas, with the

highest levels in cortex, hippocampus, and cerebellum; Al receptors are also present in

many other tissues including the heart, trachea, kidney and adipose cells. In contrast, A2a

receptors are highly concentrated in dopamine-rich regions of the brain (striatum, nucleus

accumbens and tuberculum olfactorium) and are also present in platelets, liver, and


Effects of caffeine on adenosine receptors

In the CNS, adenosine decreases neuronal firing (Kospopoulos et al., 1977; Phillis

et al., 1983) and inhibits the release of excitatory neurotransmitters via actions on Al

receptors (Fredholm et al., 1980; Hollins et al., 1980). Action of A2a receptors by

adenosine decreases the affinity of dopamine binding to dopamine D2 receptors (Ferre et

al., 1991) and blocks selected post-synaptic D2 receptor mediated actions (Jin et aL,

1993). Therefore, by binding to Al I and A2a adenosine receptors and antagonizing the

actions of adenosine, CA and related MX would be expected to promote central

stimulation (Al) and increase locomotor activity (A2a) that are regulated both by

activation of dopamrninergic neurons in the nigrostriatal pathway. In addition to actions in

the CNS, CA-induced blockade of adenosine Al receptors in peripheral tissues appears to

be responsible, in part, for the cardiovascular stimulation, bronchodilation and increased

lipid metabolism. A2a-mediatied functions outside the CNS are not well documented.

Overall, results from studies in humans and rats indicate that CA is generally a

nonselective antagonist at adenosine Al and A2a receptors with relatively modest

potency (Ki > 10 gM) compared to newer synthetic xanthine analogs (Ki 10 nM)

(Fredholm et al, 1994). Although the CA metabolites TP and XN bind to Al and A2a

receptors with greater affinity than CA in most animal species, CA is usually present in

the CNS at much higher concentrations owing to its enhanced ability to cross the BBB.

Accordingly, it is likely that CA, rather than its metabolites, accounts for most of the

effects associated with CA administration. Pharmacologically active concentrations of

CA in plasma and brain range from a threshold of 5 to 10 [LM for central stimulation to

about 50 pLM in the treatment of asthma. At these drug concentrations, adenosine

receptors would be expected to be blocked by CA and related xanthines. Therefore,

competitive antagonism of the depressant effects of endogenous adenosine has been

widely believed to be the most significant mechanism for CA's wide-range of

pharmacological effects.

Inhibition of phosphodiesterase

Phosphodiesterase is one of several enzymes that break down the intracellular

second messenger cyclic adenosine monophosphate (cAMP) that is involved in numerous

signal transduction processes (Amer and Kreighbaum, 1975). Caffeine and related MX

inhibit cyclic nucleotide PDE in peripheral tissues (Butcher et al., 1962; Beavo et al.,

1970) as well as the CNS (Vernikos-Danellis and Harris, 1968; Watanabe, 1975) and

thereby cause an accumulation of cAMP within subcellular compartments and

potentiation of cAMP regulated transduction events. Initially, investigators had proposed

that the mechanism of inhibition by MX was thought to involve direct interaction with

PDE (Daly, 1993). However, since PDE inhibition requires an extremely high drug

concentration (0.2 to 1 mM), this appears unlikely owing to the fact that MX exhibit most

pharmacological actions at much lower levels in vivo (Cardinali, 1980; Wachtel, 1982). A

link between PDE inhibition and CA pharmacological activity has not been conclusively

established. This is especially so in the CNS where blockade of adenosine Al receptors

also increases cAMP concentration. So far, there are three systems where PDE inhibition

has been proved to be involved in the pharmacological activities of CA and TP: the

cardiovascular system, the respiratory system (bronchial relaxation) and the

muscular/adipose tissues (increased lipolysis)(Daly, 1993).

Mobilization of intracellular calcium

The effect of MX on mobilization of intracellular calcium was first demonstrated

in skeletal muscle and later in mammalian cardiac muscle (Guthrie and Nayler, 1967).

Caffeine, at a concentration of 1 to 2 mM lowers the excitability threshold and prolongs

the duration of the active period for muscle contraction by promoting translocation of

calcium through the plasma membrane and sarcoplasmic reticulum (Bianchi, 1961, 1968,

1975). Direct interaction of MX with calcium channels has been demonstrated in the

sarcoplasmic reticulum. (Rubtsov and Murphy, 1988). In the CNS, increased neuronal

excitability (Kuba and Nishi, 1976) and enhanced neurotransmitter release (Poisner,

1973) have been linked to the mobilization of calcium from intracellular storage sites by

high CA concentrations (mM). However, since a minimum concentration of 250 uLM of

CA is the threshold necessary to produce detectable effects on calcium shift from

intracellular storage sites (Sandow and Brust, 1966; Guthrie and Nayler, 1967), it seems

unlikely that calcium mobilization represents an essential step for many of CA's actions

aside from possible toxic effects such as the production of seizures. Nevertheless,

pharmacologically relevant concentrations of CA can increase intracellular calcium levels

in skeletal muscle indirectly through various pathways including increases in fatty acyl

CoA levels, adenosine Al receptor blockade and enhanced release of epinephrine and

cortisol (Spriet et al., 1992; Sinclair and Geiger, 2000). These results underscore the

importance of pharmacological effects secondary to antagonistic binding at adenosine


Interactions with dopaminergic and benzodiazepine receptors

Results from anatomical and functional studies indicate a possible involvement of

dopaminergic systems in the central effects of CA, especially those effects that influence

motor control and locomotor activity (Ferre et al., 2000). The ability of CA and related

MX to increase dopamine receptor-mediated rotational behavior in rats with unilateral

nigrostriatal pathway lesions was first reported in 1976 (Fredholm, 1976) and has since

been confirmed and elaborated on by other investigators (Ferre et al., 1992; Daly, 1993;

Ongini and Fredholm, 1996). Selective Al receptors agonists reduce the high affinity

binding of agonist to dopamine Dl receptors, whereas A2a receptor activation reduces

agonist binding affinity at dopamine D2 receptors. In addition CA has been reported to

cause dose-dependent increases in extracellular dopamine levels in the striatum of mice

(Morgan and Vestal, 1989; Okada et al., 1997). Evidence has been obtained for the

possible formation of receptor heterodimers between adenosine Al and dopamine Dl as

well as adenosine A2a and dopamine D2 receptors (Ferre et al., 2000). All these findings

suggest that interactions and probably receptor subtype-specific interactions occur

between adenosine and dopamine receptors. Nevertheless, the specific sites of adenosine

receptor antagonism that are linked to CA-induced motor stimulation may involve

transmitters other than dopamine, including GABA, serotonin and acetylcholine-

containing pathways (Daly, 1993).

Caffeine has anxiogenic effects in humans (Hughes et al., 1991), so the

interaction of CA with CNS sites that bind the anxiolytic benzodiazepine drugs is not

unexpected. Caffeine can antagonize several effects of benzodiazepines including central

muscle relaxant effect and anxiolytic properties (Pole et al., 1981), as well as sedation

and impairment of motor performance (Roache and Griffiths, 1987). Conversely,

benzodiazepines counteract the locomotor stimulation produced by CA (DeAngelis et al.,

1982) and prevent CA-induced seizures with a rank order of potency that is

commensurate with their binding affinities at benzodiazepine binding sites (Czuczwar et

aL, 1985). However, the direct affinity for CA at benzodiazepine binding sites is very low

(Marangos et al., 1979; Boulenger et al., 1982) and coincides with in vivo concentrations

of CA that produce seizures and convulsions. Therefore, binding of CA to

benzodiazepine receptors is not likely to contribute in any substantial way to the overall

action of CA at behaviorally effective doses.

Caffeine Use and Athletic Performance

Caffeine has long been viewed as a performance-enhancing agent. It has been

widely used by athletes in an effort to gain advantages through its ergogenic properties.

While studies showed that CA effectively increases athletic performance in endurance

events (Sinclair and Geiger, 2000), the impact on short burst-type athletic competition is

not well documented (Graham et al., 1994; Doherty, 1998). Nevertheless, it has been

suggested that the wide spectrum of pharmacological actions by CA and its metabolites,

which involve both primary and secondary effects on neurotransmitters, energy

metabolism and numerous other process, presents a significant potential to enhance

athletic performance in subtle yet effective ways. In view of these beliefs, it seems

unlikely that CA use by human athletes will decline any time soon because of its

inexpensive cost, ready availability, safety, social acceptance and legal status. The

International Olympic Committee (IOC) and the National Collegiate Athletic Association

(NCAA) regard CA as a "restricted substance" with an allowable limit of 12 tg/mL in

urine. In the animal racing industry, CA has been designated as a class 2 agent by the

Association of Racing Commissioners International (ARCI) and is banned in racing

animals. The concept of allowable CA levels in human athletes has been one intensive

debate in regulatory bodies around the world. For instance, while a 12 g.g/mL limit is

considered too low in Australia because the level may be easily exceeded with

consumption of 3 to 5 cups of coffee (Birkett and Miners, 1991), the level is considered

too high in Canada (Canadian Center for Ethics in sport, 1997) because some

performance enhancing properties occur at levels below the current limit (12 p.g/mL)

(Wagner, 1990; Graham and Spriet, 1991). To resolve such disagreements, one must keep

in mind that the goal of drug-testing in sports is to protect the health of athletes and to

enforce the fairness of the competition by preventing any possible advantages that may

be gained from consumption of foreign substances. Using this guideline, it seems obvious

that the allowable levels of CA in urine might need to be further restricted since CA has

been shown to enhance endurance performance at levels well below the 12 u.g/mL (= 60

tM) limit. In addition, it seems reasonable that pharmacologically active CA metabolites

(TP, XN and TB) should also be included in the test limit even though such testing may

impact athletes who consume coffee or tea on a daily basis. Adding to the confusion,

current IOC rules ban any substance (natural or synthetic) that elevates ACTH or GH

(Sinclair and Geiger, 2000) and CA exhibits both actions (Lovallo et al., 1996; Lin et al.,


The regulatory issue of CA in racing animals is quite different from that in

humans because CA is not normally consumed by animals. Nevertheless, in view of the

general acceptance of CA in humans and the easy access to this substance in numerous

products consumed by humans, the complete prohibition against any detectable CA in

racing animals has spawned requests for further studies to verify the ability and/or extent

to which CA enhances the performance of racing animals. In the case of horses, which

comprise probably the largest monetary share of the animal racing industry, studies of

CA have been very limited (Kurosawa et al., 1999) and largely confined to the

pharmacokinetic properties of this compound (Aramaki et al., 1991, 1995, 1996; Peck,

1997). While pharmacokinetic studies provide important information with regard to

threshold levels, they have shed little insight into the pharmacodynamic actions that are

important in relation to possible performance-enhancing effects. Therefore, combined

studies of pharmacokinetics and pharmacodynamic actions are needed. More specifically,

the pharmacokinetics of CA in CNS and the ergogenic effects of CA in the horse at both

behavioral and cellular level warrant further investigation. Such studies would be

important in establishing the relationship between CA concentrations at potential sites of

action and CA effects associated with performance. The best-characterized mechanism of

actions of CA in the CNS, the antagonism of adenosine receptors, needs to be evaluated

in horse brain tissues.

To achieve such goals, a system capable of readily accessing chemical events in

the CNS is crucial as is its incorporation in a model that allows concurrent assessment of

drug levels and CNS stimulation effects. The combination of the two key features was

not possible until the application of modem microdialysis technique to free moving

animals became available. Because of the importance of this technique, the second half of

this chapter introduces and critically discusses the technology and application of


Introduction to Microdialysis Technique

The analysis of endogenous or exogenous compounds in the body has

traditionally involved chemical analysis of blood, or measurements in dissected tissues.

However, such measures have limited utility insofar as the blood is a distant reflection of

events that take place in cells or tissues; and dissected tissue presents a static picture of

dynamic chemical events within tissue compartments. While chemical communication

between cells occurs in the extracellular fluid (ECF), this compartment is frequently not

considered due to its physical inaccessibility.

Several methods have been used in order to monitor chemical changes in the

extracellular environment of tissues. In 1961, Gaddum described a push-pull perfusion

method, which involved the stereotaxic insertion of a small-diameter cannula into a

selected area of the rodent brain (Gaddum, 1961). The cannula consisted of two

concentric stainless steel tubes in which a smaller push cannula was inserted into an outer

pull cannula. By infusing an artificial solution via the inner cannula, small amount of

ECF could be withdrawn simultaneously without causing any net change in ECF volume.

For over 20 years, the push-pull technique was used routinely to measure in vivo levels of

neurotransmitters and other endogenous factors in the brain (Philippu, 1984; Gardner et

al., 1993; Myers et al., 1997). In 1966, Bito and coworkers first implanted a "dialysis

sac" into the subcutaneous tissue of the neck and into the parenchyma of the cerebral

hemispheres of dogs (Bito et al., 1966). These experiments introduced the concept of a

"compartment" surrounded by a semi-permeable dialysis membrane that would allow

equilibration of small molecules from the extracellular environment. In 1972, Delgado

and colleagues (Delgado et al., 1972) developed an improved push-pull cannula called

"dialytrode" in which a semi-permeable membrane was used to cover the cannula tip and

thereby reduce tissue damage and avoid microbial contamination at the site of perfusion.

Even though the dialytrode technique was a novel improvement from the push-pull

cannula and was adopted in neuroscience research (Kovacs et al., 1976; Delgado et al.,

1984), it did not find extensive application because dialytrode was soon replaced by a

"dialysis membrane" that consists of a hollow fiber. This replacement marked the dawn

of the modem microdialysis technique (Ungerstedt, 1984; Ungerstedt and Hallstrom,

1987). Because of its relative ease of use, microdialysis has become in most cases the

technique of choice for in vivo analysis of chemical events in the extracellular tissue

spaces. The number of scientific articles reporting the use of microdialysis has steadily

increased from 250 in 1991 to 750 in 1996 and more than 3600 as of 1998 (Chaurasia,


Principals of Microdialysis

The idea of microdialysis is to mimic the passive function of capillary blood

vessels by perfusing a thin dialysis tube implanted into the tissue. The term dialysis refers

to the bi-directional passage of small molecules and water through a semi-permeable

membrane where the process is driven by the concentration gradients of compounds. By

virtue of bi-directional diffusion across the dialysis membrane, both endogenous and

exogenous substances can be delivered into or recovered from the ECF. Microdialysis

experiments are carried out by implanting a small diameter probe, consisting of a

semipermeable hollow fiber membrane, into target tissue or biological matrix of interest

and perfusing the probe with a suitable fluid (the perfusate) at a very slow flow rate. The

perfusate is an aqueous solution that must closely match the (ionic) composition of the

ECF surrounding the probe in order to prevent changes in ECF composition. Low

molecular weight substances could diffuse from the perfusate into ECF or from ECF into

the perfusate depending on the prevailing concentration gradient. Substances that pass

through the membrane are collected (the dialysate) and carried out of the body for

subsequent analysis.

When compared with more conventional sampling techniques, microdialysis

offers several distinct advantages. The probe can be implanted in almost any tissue or

organ where only lx lx 3 mm3 tissue is needed for placement of the probe. Samples can

be obtained with less stress to the animal and therefore do not significantly change the

physiological state of the animal, and samples can be collected simultaneously from

multiple organs. Such advantages have enabled investigators to compare pharmacokinetic

parameters for drugs within multiple compartments. In addition, microdialysis samples

provide a more consistent and accurate measure of drug levels since they are collected

continuously and represent a time-averaged measure over the collection interval rather

than separate measures at discrete time points. Furthermore, since there is virtually no net

removal of body fluid, it is feasible to obtain multiple samples over a prolonged period

from each experimental subject and thus reduce the number of experimental animals

needed for pharmacokinetic studies. Additionally because of the less physiological

derangements involved, crossover experiments can be done within a single animal. With

the microdialysis principle providing free or non-protein-bound form of the drug, a

measure of drug binding to proteins in vivo could be obtained when used in combination

with conventional sampling techniques. A much less contaminated sample that is devoid

of protein and cells is collected thereby reducing interference with drug analysis

techniques. Moreover, in view of the adaptability of this technique to free-moving

animals, microdialysis allows for simultaneous monitoring of drug levels and

physiological or behavioral changes produced by the drug.

Microdialysis Systems

A basic microdialysis system consists of a microdialysis probe, a perfusion pump,

inlet and outlet tubing, and a (refrigerated) microfraction collector. The microdialysis

probe can be manufactured in the laboratory or purchased commercially. The perfusion

pump should be able to provide an exact and pulse-free flow rate in the nL/min and

utL/min range, while the microfraction collector should be able to collect volumes exactly

according to pre-set volumes at pre-set intervals. The length and inner diameter of the

outlet tubing is chosen to minimize longitudinal diffusion of the dialysate and to prevent

hydrostatic pressure build-up across the probe membrane.

Perfusion media used in microdialysis experiments vary widely in composition

and pH (Benveniste and Huttemeier, 1990). Ideally the composition, ion strength,

osmotic value and pH of the perfusion solution should be as close as possible to those of

the ECF of the dialyzed tissue. In CNS studies, non-physiological hypotonicc) perfusion

medium has been reported resulting in a substantial increase of the BBB permeability (De

Lange, et al., 1994). In practice, perfusion fluid used for microdialysis differs among

individual laboratories incorporating fluids such as normal saline, buffered Ringer's

solution or artificial CSF (Benveniste and Huttemeier, 1990, de Lange et al., 2000).

Selected deviations in ion composition or addition of chemical agents to the perfusion

medium have been used for different purposes including alteration of free calcium

concentration (modulate transmitter release) (Augustine et al., 1987; Moghaddam and

Bunney, 1989; Osborne etal., 1990; Timmerman and Westerink, 1991) or changes in

glucose levels (evaluate glucose-dependent metabolic processes) (Ronne-Engstrom et al.,

1995). In some cases, proteins have been added to the perfusion medium to prevent

sticking of the drug to the microdialysis probe and tubing connections (Kendrick, 1989;

Levine and Powell, 1989; Maidment et al., 1989). In addition, drugs may be included in

perfusion medium and introduced locally into tissues in order to prevent cellular transport

or inhibit enzymes that degrade analytes of interest. However, the impact of these

experimental manipulations must be viewed with caution insofar as tissue responses may

have been altered or further influenced by drug interactions (Cuadra et al., 1994; De Boer

et al., 1996). Precautions have to be taken on the temperature of perfusion medium. Most

investigators use a perfusion medium at room temperature before entering the probe. As a

result a temperature gradient exists between the probe and its environment and may

consequently affect the results. This is especially important for studies of tissues under

pathological circumstances where abnormal tissue may lose its capability to compensate

for temperature changes (De Lange et al., 1994). Therefore, it is recommended to

perform all microdialysis experiments with perfusion fluids at body temperature.

Tubing connections ideally should have no interaction with the drug as this may

have a profound effect on the relation between the concentration of the drug found in the

dialysate and the ECF. This should also hold true to the probe membrane. The inner

diameters of the tubing connections are of importance with respect to build-up of fluid

pressure and thereby fluid loss over the semi-permeable membrane. This may be

prevented by using inlet tubing (from perfusion pump to the probe) with an inner

diameter smaller than that of the probe itself, and an outlet tubing (from probe to

collection site) with an inner diameter being larger than that of the probe. Although there

is no general rule regarding the maximum applicable length of tubing, the minimum

length of tubing should be considered and the outlet tubing normally should not exceed 2


Microdialysis Probe and Recovery

The microdialysis probe is the core part of the microdialysis system. A basic

probe contains a dialysis membrane attached to an inlet and an outlet tube that are usually

threaded through another tubing. All microdialysis probe components are made of

biocompatible materials. Typically, the hollow fiber semi-permeable dialysis membrane

is made of a polycarbonate-ether copolymer, regenerated cellulose, cellulose acetate,

cuprophane, or polyacrylonitrile (PAN), with molecular weight cutoffs ranging from

5000-50,000 Da. More recently, a polyether sulfonate (PES) membrane has been

introduced with a greater molecular weight cutoff of 100,000 Das (Elmaquist and

Sawchuk, 1997) which facilitates microdialysis sampling of environmental toxicants that

usually have long carbon chains and larger molecular size. The outer diameter (O.D.) for

most probes ranges from 150 to 500 Jtm, with dialysis membrane lengths from 1 to 10

mm. The narrow-bore tubing attached to the inlet and outlet of the probe is made of inert

materials such as fused silica, Teflon, or polyethylether ketone. The dead volume of

tubing can vary from 0.044 to 0.12 uL/cm in commercially available probes. The probes

generally tolerate ethylene oxide gas sterilization well without damage to the membrane.

This is not so for thermal sterilization.

There are two basic probe geometries for in vivo microdialysis. These are referred

to as parallel and serial type probes. Choice of probe geometry is based on the site of

implantation and its surgical accessibility. Parallel perfusion probes are constructed of

rigid (concentric type) or flexible (side-by-side type) cannula material and are usually

shorter in membrane length. Probes with rigid cannulae are especially suited for intra-

cerebral sampling whereas the flexible cannulae are advantageous for venous or soft

tissue sampling. Serial perfusion probes include linear and loop styles and usually are

made with longer membrane for better recovery. They are typically used for soft

peripheral tissue such as skin, muscle, tumor and liver or fluids like blood and bile.

Modifications from the two basic configurations have also been developed to facilitate

tissue-specific sampling (Scott and Lunte, 1993; Kanthan et al., 1995; Marsala et al.,

1995; Khan et al., 1996; Evrard et al., 1996; Dempsey et al., 1997).

By its very natures, microdialysis is a dynamic process whereby substances are

continuously moving to and from the sampling site via diffusion across the probe. As the

dialysis procedure is usually performed under a steady state condition rather than a

complete equilibrium with analyte concentration in ECF, consequently the concentrations

of the drug in the dialysate reflect only a fraction of the drug concentrations in the ECF.

The term recovery is used to describe the relation between concentrations of the drug in

the ECF and those in the dialysate. Absolute recovery (or mass recovery) is defined as the

amount of drug that is extracted (or delivered) by the dialysate as a function of time.

Relative recovery (or concentration recovery) refers to the fraction or percent of analyte

that is recovered from the extracellular compartment (in vivo) or sample matrix (in vitro).

In practical terms, relative recovery is defined as the concentration of analyte in dialysate

divided by its uniform concentration in the ECF or sample matrix, and is commonly

expressed as a percent value. In general, the term recovery refers to the relative recovery.

Several factors may impact recovery of analyte and must be considered when one

attempts to carry out quantitative measures of analyte concentration with this technique.

Probe recovery is most significantly influenced by perfusion flow rate, perfusate

composition, characteristics of the probe (geometry, membrane surface area, and pore

size), and characteristics of the analyte (size, charge and relative hydrophobicity). Other

factors that could influence recovery include tissue toturosity and temperature. Various in

vitro experiments have been performed to examine the relationship between flow-rate

and in vitro recovery. The results indicated that in vitro recovery is inversely dependent

on the flow-rate (Johnson and Justice, 1983; Jacobson and Hamberger, 1985; Hemrnandez

et al., 1986; Tossman and Ungerstedt, 1986; Lindefors et al., 1989; Nakamura et al.,

1990). A higher flow-rate will increase the fluid pressure inside the probe, which may

result in net transport of perfusate across the dialysis membrane and thereby counteract

the diffusion of the drug into the dialysates. The length (Kuipers and Korf, 1994) and the

sequence (Ruggeri et al., 1990) of dimensions of the inlet and outlet tubing will also

greatly affect the fluid pressure gradients that may ultimately alter the recovery. It is

recommended to use a sequence of diameters of tubing that gradually increase from the

perfusion pump to the collection site and to check flow-rate with and without the probe

interconnected. Perfusate composition, especially a difference between the osmotic

values of the perfusate and that in the ECF, will affect recovery as water movement will

either add on or counteract diffusion of the drug across probe membrane (Borg and

Stahle, 1999). An increase of the dialysis surface (length), as often seen in linear type

probes, will increase the recovery (Johnson and Justice, 1983; Lindefors et al., 1989).

However, this increase will only be linear with small surfaces because, at larger surfaces,

the increase in recovery will start to lag behind the increase in surface because the

concentration difference between ECF traversing along the probe membrane will

gradually diminish with the longer length of the membrane. Since the net movement of

substances across a probe membrane depends on the physical and chemical properties of

the analyte, the diffusional properties of the analyte as well as the geometry of the probe

and membrane pore size (Benveniste and Huttemeier, 1990; Parry et al., 1990;

Benveniste and Hansen, 1991) will affect recovery. It is a general rule that higher

recovery usually associates with analytes of smaller molecular weight and higher

hydrophilicity, and probe membrane with larger pore size. A higher body or perfusate

temperature will also increase recovery.

While all parameters that influence in vitro recovery influence in vivo recovery it

must be remembered that tissue characteristics will play an important role and may

ultimately determine the recovery. This is because the effective diffusion of the drug

through the ECF of a tissue will be affected by uptake into cells, metabolic conversion

rate, and active transport across membranes, extent of tissue vascularization and blood

flow through the tissue. In other words, the implications of some parameters that would

affect recovery in vitro may be different from in vivo or even different between two

tissues. For example, in solid tissues, effective diffusion through tissue membrane to the

ECF determines the recovery of the microdialysis probe (Bungay et al, 1990; Morrison

et al., 1991) while diffusion through the dialysis membrane is rate limiting when

transport occurs in a rapid flowing systems (like blood). Consequently, with solid tissues

calibration performed in vitro may not be valid in vivo (Hsiao et al., 1990) but for

flowing fluid systems, an in vitro recovery could provide very accurate estimation of in

vivo recovery values if perfusion flow-rate and temperature are well controlled. There are

physiological responses to probe implantation (for example, clot formation around an

intravenously implanted probe) which might occur and result in erroneous recovery

estimation due to changes in the tissue environment. One factor that does not affect

recovery is analyte concentration (Sandberg and Indstrom, 1983; Ungerstedt, 1984;

Muller et al., 1995). Since by nature the recovery process depends on simple diffusion,

the rate of transfer across the membrane should be equivalent at all concentrations under

constant experimental conditions. In summary, there are numerous factors that could

potentially affect recovery, and the rule of thumb is to keep all experimental conditions

(variables) as uniform as possible. An optimal recovery should balance between

perfusion flow rate and sampling interval (temporal resolution). Most investigators use a

flow-rate that ranges from 0.1 to 5 tUL/min with 2 uL/min being most popular in view of

higher flow rates may remove more endogenous compounds and thereby more easily

locally distort normal physiology (Gonzalez-Mora et al., 1991). Nevertheless, flow-rates

as high as 10 pL/min were not uncommon.

Quantitation of Microdialysis Data

Validation of probe recovery

Determination of probe recovery is a critical issue for any microdialysis

experiment because the "true value" of the concentrations of interested analytes can not

be converted from the dialysate concentration without a known validated and accurate

value for recovery. A major concern in the increasing adoption of microdialysis sampling

for pharmaceutical and pharmacological studies is the calibration of the microdialysis

probes. Microdialysis is generally performed under nonequilibrium conditions where the

recovery of a compound depends on many parameters as described earlier in this chapter.

It is not always clear whether a high or low recovery is preferable. High recoveries are

favored from the analytical perspective because higher concentrations are obtained. A

high recovery also results in less depletion of the analyte and other endogenous

compound concentrations around the probe because the situation is closer to equilibrium.

However, high recovery is typically achieved either using very slow perfusion rate or

long dialysis membranes. These manipulations have the drawback of loss of temporal

resolution (slow perfusion rate) and loss of spatial resolution (large sampling surface).

Low recoveries, on the other hand, result in less concentrated samples for analysis and a

greater decrease in the analyte concentration around the microdialysis probe. However,

these limitations may be offset by the fact that higher perfusion rates and shorter dialysis

membranes can be used. In vitro recovery is useful in determining an optimum flow rate

and for examining probe-to-probe variability. It is usually performed before in vivo

experiments to determine the efficiency of recovery for particular analytes and for

evaluating probes integrity after experiments. However, in vitro recovery, which is

usually performed in aqueous solutions, tends to overestimate the in vivo recovery from

tissue and does not account for the nature of the tissue to be sampled and its interactions

with the analyze also affect the recovery. Therefore, converting in vivo results from in

vitro recovery usually underestimate the true value of analytes in the tissue and are

considered inappropriate (Amberg et al., 1989; Benveniste et al., 1989; Bungay et al.,

1990; Stable et al., 1990). Development of more accurate in vivo calibration methods for

probe recovery has occurred in recent years (Bungay et al., 1990; Larsson 1991; Stahle et

al., 1991b; Menacherry et al., 1992; Le Quellec et al., 1995). While none of these

methods is clearly more beneficial than the other, satisfactory results could usually be

obtained by carefully evaluating the method of choice or by modification of the existing

methods. Nevertheless, for many experiments, accurate calibration of the microdialysis

probe is not necessary. For example, if the desired information is the relative change in

concentration introduced by some experimental manipulation, only the concentration

independence and stability of the recovery need be known. On the other hand, if a

relative distribution of a compound to various sites is the information desired, then the

behavior of each probe used must be normalized against the other probes. Only when an

absolute concentration is needed must each probe be calibrated for an accurate in vivo

recovery. As this is often the most difficult step in a microdialysis experiment, careful

experimental design should be practiced. If the desired information can be obtained from

relative rather than absolute concentrations, the calibration issue is greatly simplified. A

number of approaches to the calibration of microdialysis probes in vivo have been

described. The two most common methods involve adding an internal standard to the

perfusate, which is commonly known as "retrodialysis," or estimating the equilibrium

condition by alternating varying concentrations of analyte to the perfusate, which is

known as the method of zero net flux (ZNF). Alternatively, very slow perfusion rates can

be employed, often circumventing the need for calibration.

Slow perfusion rate method is based on the assumption that equilibrium exists

between sampling environment and perfusate at very slow flow rate. It has been

demonstrated that at perfusion rates slower than 50 nL/min, recovery is greater than 95%

for compounds with molecular weights smaller than 500 Da (Menacherry et al., 1992).

Therefore, at such a slow perfusion rate, the error, introduced by assuming 100%

recovery, is insignificant. The drawback to this approach is the long sampling intervals

required to collect sufficient sample volume for analysis. In the retrodialysis method, an

internal standard is added to the perfusate and the rate of delivery of this compound to the

tissue is measured during microdialysis sampling (Larsson 1991; Wang et al., 1993). The

internal standard should match the characteristics of the drug as close as possible so that

the concentration loss of the internal standard will predict the concentration gain

(recovery) of the drug. The assumption is made that the recovery ofanalyte is equal to

the delivery of the internal standard. The advantage of this method is convenience and

time saving, however, the assumption that effective diffusion of the internal standard and

drug is equal may not be realistic (Stahle, 1991a,b), and the similarity between internal

standard and analyte of interest may introduce unwanted interactions between the two in

the sampling area. Alternatively, the drug itself may be added to the perfusate and its in

vivo loss may be used as a measure for in vivo recovery (i.e. in vivo delivery method). In

principle this measurement can be conducted before the start of the real experiment at the

time when there is no drug present in the body. This method is simple, convenient, and

timesaving, however, the assumption of this approach that the probe provides bi-

directional movement of the same analyte at equal efficiency should be validated at least

in vitro (Ault et al., 1994; Davies, 1995). This is because tissue specific asymptotic

profiles in recovery that result under increasing concentration gradient conditions have

been reported (LeQuellec et al., 1994). Subsequently, as a modification ofretrodialysis

method, a combination of using internal standard and analyte (the drug) of interest in

retrodialysis has been recently developed (Bouw et al., 1998). The relative loss for both

compounds is first determined and related to each other by calculation of their ratio. The

second step involves the washout period for the drug by perfusion of the probe with the

internal standard only. The third step of the study involves the actual administration of

the drug to the body, and measuring the concentration of the drug in the dialysate

samples. During this third stage, internal standard is still present in the perfusion solution

and it's relative loss is taken as the recovery (or relative gain) of the drug for each sample

interval, using the loss ratio as determined in the first step of the study. The idea here is to

further scrutinize the difference in recovery between an internal standard and the analyte

of interest.

The other most used in vivo method for probe validation is ZNF method

(Lonnroth et al., 1987). This method involves consecutive perfusion of the microdialysis

probe with different concentrations of the analyte at a steady-state concentration in the

tissue. A plot of the change in perfusate concentration versus the initial perfusate

concentration should be linear with the slope being the inverse of the probe recovery and

the y-intercept being the point of ZNF; that is, the concentration of the analyte in the

tissue. This method yields generally higher accuracy than other methods but is very time-

consuming and the tissue concentration of the analyte must remain at steady state during

the entire measuring process, which required precise dose calculation. In addition, it is

inadequate for monitoring drug concentrations and probe recovery as a function of time.

To estimate in vivo recovery as a function of time, Olson and Justice (Olson and Justice,

1993) presented an extended version of the ZNF method, which is called the "dynamic-

zero-net-flux" (DZNF) method. Instead of serial perfusion of individual animals with

different concentrations via the probe, a group of animals are continuously perfused with

one selected perfusion concentration. Different groups receive different concentrations

and the results are combined at each time point. Regression of the mean data points of the

different groups at a particular point of time will give the actual ECF concentration with

the associated in vivo recovery value at that time. These investigators showed that in vivo

recovery might change during the course of an experiment. This may have important

implications for the interpretation of results obtained in earlier microdialysis studies,

which might have been erroneous. Although this is a powerful experimental setup, more

experimental animals are needed, which in part reduces the advantage of minimizing the

use of living experimental animals by the microdialysis technique.

Among other methods, zero net flow method refers to the strategy that the

dialysate concentration is measured at various perfusion flow rates, plotted against flow

rate and extrapolated to zero. The concentration at zero flow is presumed to equal the

extracellular concentration, assuming that at zero flow the system is at equilibrium. As in

ZNF method, this method is time consuming and the validity of this method also relies on

a steady-state of analyte concentration through the whole sampling period (Jacobson et

at., 1985). In summary, the choice of appropriate in vivo probe calibration methods is

largely dependent on the purpose of the study. For studies that require information only

on relative concentration changes of the analytes, retrodialysis could be useful. For

studies in hydrodynamic system (vascular system) or where only approximation of

analyte concentration was required, in vitro calibration may be sufficient if the perfusion

flow rate and sample temperature are well controlled. For those accurate in vivo

concentration of analytes is required, approaches such as very low flow rate, ZNF or

retrodialysis should be employed.

Analytical considerations

When microdialysis sampling is applied in vivo, the subject, the sampling interval

and the analytical system must be considered in relation to each other (Wages et al.,

1986). With a 10-minute collection interval, sample volumes of 1 to 50 jl are obtained;

sample evaporation can become an important concern (Deterding et al., 1992; Hogan et

al., 1994). On-line analysis could be applied to minimize sample loss (Chen and Lunte,

1995). However, for on-line sample analysis the analysis time should be shorter than the

sample collection time thus requiring an analytical procedure with shorter sample run

time. The dialysate collected by in vivo microdialysis can be analyzed by any available

analytical procedure, provided that it is able to deal with small sample volumes

(Kissinger and Shoup, 1990). Mostly, reversed phase high-pressure liquid

chromatography combined with ultraviolet, fluorescence, electrochemical detection

(Cheng and Kuo, 1995), or mass spectroscopy, are used because these may provide

sensitive analysis of small volumes. If drugs are present in the dialysate at very low

concentrations, as is often the case, more sensitive analytical techniques are needed.

Then, microbore liquid chromatography, capillary liquid chromatography combined with

on-column focusing, and capillary electrophoreses with laser-induced-fluoresce detection

become more appropriate (Tellez et al., 1992; Robert et al., 1995). These techniques can

handle extremely low sample volumes (nL), which means that a lower perfusate flow-rate

can be used to increase the recovery and/or shorter collection intervals can be adopted to

increase the temporal resolution.

Limitations of Microdialysis

Although the small size of the probe, mostly 300 p.m O.D., causes minimal

disturbances within the tissue, implantation of a probe is still invasive and effects on the

target tissue should be determined in order to prevent the possibility of generating

erroneous results. Probe insertion can cause direct tissue trauma, and secondary tissue

responses including changes in blood circulation and foreign body reactions (Anderson et

al., 1994). Local tissue reactivity (flare, erythema) is the most prominent short-term

effect, whereas foreign body reaction (neutrophil, macrophage infiltration) is more

pronounced in the longer term. Samples collected in the immediate post-insertion period

may not reflect basal conditions due to substances released secondary to tissue injury,

changes in blood flow, or changes in the dialysis efficiency of the probe in traumatized

tissue. Therefore, immediately after insertion, an equilibration period is usually allowed

before beginning sampling. Peripheral tissues including dermis, muscle, tumor, and liver

have been examined on their response to probe implantation (Ault et al., 1994; Palsmeier

and Lunte, 1994; Davies and Lunte, 1995). In these studies, initial infiltration of

neutrophils was found in dermis, muscle and liver followed by the presence of

macrophages. In tumors, little or no inflammatory response was found up to 72 hours

after probe implantation. In the evaluation of tissue trauma following intracerebral

implantation of a microdialysis probe, initial formation of eicosanoids, local disturbances

in cerebral blood flow and glucose metabolism were present (Benveniste et al., 1987;

Yergey and Heyes, 1990). These changes were more or less normalized 1 day after

surgery. Histological evaluation revealed that glial reactions (gliosis) usually started 2 or

3 days after implantation of the probe and reactions were confined mainly to a very small

region around the probe (Benveniste and Diemer, 1987; Shuaib et al., 1990). These

results indicate that the optimal period or interval to perform (repeated) intracerebral

microdialysis experiments, in general, lies between 1 and 2 days after implantation of the

probe or after recovery from early tissue reactions, and before the start of long-term

tissue reactions (De Lange et al., 1994, 1995). With respect to anesthesia, the use of

anesthetics may interfere with physiological processes (Claassen, 1994); in order to

reduce the effects of anesthesia, a guide cannula can be used through which the probe can

be inserted after animal had recovered from the surgery (Hamilton et al., 1992; Drijfhout

etal., 1995).

The majority of applications of microdialysis thus far include the brain. In

pharmacokinetic studies on the distribution of drugs into the brain, transport across the

BBB is an essential factor. With the introduction of a microdialysis probe it is important

that the BBB is intact at time of measurement. The evaluation of the integrity of BBB is

commonly accomplished by employing markers that do not normally across the barrier

(Westergren et al., 1995; de Lange et al., 1997), followed by the determination of the

marker concentration in the brain. The influence of tissue damage to BBB integrity

apparently would be more significant for small lab animals than for large animals

because, relatively, a larger percentage of tissue area is affected by the procedure.

Quantitation or calibration of microdialysis probes and the challenge of analyzing

the small volumes typical of microdialysis sampling are issues that must continue to be

improved. For better recovery, increased membrane length and molecular cutoff, longer

sampling period and decreased perfusion flow rate are desirable; but these factors work

against temporal resolution and higher risk of tissue damage and unwanted endogenouss)

analytes collection. Balance should be made among the above factors to reach optimum

recovery in order to meet analytical sensitivity. Furthermore, probe recovery also tends to

decrease with time and varies among different tissues and chemical substances, it is

therefore recommended to evaluate probe recovery in each tissue before and after the

experiment. This may be time-consuming and partially counteract the advantage of


Unlike pharmacokinetic studies of drugs where distribution and elimination

occurs at time frames of minutes to hours, neuronal stimulation on transmitter release

usually occur at sub-second time scales. This has posted great difficulties for

microdialysis to reach optimal immediate temporal resolution in view of the fact that

minimum sample volume requirement for traditional analytical methodology simply can

not be meet within such short period of time by microdialysis. Recent advances in the

detection of very low concentrations of certain transmitters with capillary

electrophoresis/laser induced fluorescence detection have permitted a considerable

reduction in the sampling intervals from 30 seconds to 5 seconds (Thompson et al.,

1999). It is hoped that future methodological and technical developments will overcome

some of these limitations. In addition to temporal resolution, spatial resolution has also

been a noticeable concern. Spatial resolution refers to how accurate microdialysis probe

can target only the desired cells or tissues. Although currently only 1 to 3 mm x 300 gim

tissue area is required for probe implantation, it is apparent that while less problematic

for larger animals, spatial resolution will consistently be a significant concern for smaller

tissue regions with high heterogenicity like rat or mice brain.

Application of Microdialysis

Originally designed for neuronal studies, microdialysis technique has since gained

increasing applications in not only the various aspects of neurology (Adell et al., 1998),

but also in pharmacology, pharmacokinetics (Scott et al., 1991; Alonso et al., 1995;

Fettweis and Borlak, 1996; Elmqnist and Sawchuk, 1997; Hansen et al., 1997), drug

metabolism, toxicology, electrophysiology, endocrinology and behavioral/psychology

(Abercrombie et al., 1988; Sarter and Bruno, 1999). It also has become a regular

technique for study of drug-protein bindings (Herrera et aL., 1990; Ekblom et al., 1992;

Sarre et al., 1992; Quellec et al., 1994) and readily employed for delivery of drugs to

specific tissues (Hughes et al., 1996; Boschi and Scherrmann, 2000; Muller, 2000). The

tissue sites where the probes have been implanted have expanded from various regions of

the brain to specific tissues, cavities and organs of interest including blood/bile (Telting-

Diaz et al., 1992; Scott and Lunte, 1993), skin (Ault et al., 1992; Schnetz and Fartasch,

2001), muscle (Deleu et al., 1994; Henriksson, 1999), adipose tissue (Stahle et al., 1991b;

Jansson et al., 1994; Lonnroth, 1997), heart (Mertes et al., 1994; Obata et al., 1994;

Timoshin et aL, 1994), lung (Ingvast-Larsson, 1992), liver (Davies et al., 1993; Van

Belle et al., 1995), kidney (Baranowski and Westenfelder, 1994), eyes (Waga and

Ehinger, 1995) and even tumors (Palsmeier and Lunte, 1994). With carefully evaluated

surgical procedure and probe modification, the potential for the application of

microdialysis technique to any tissues and organs is feasible. The ability ofmicrodialysis

to reach deep tissues and organs for continuous sampling of protein-free samples over a

prolonged period of time has special implications in the pharmacokinetic studies in which

drug adsorption, distribution, metabolism and elimination were the focus. Aside from the

established success in studies ofneurosciences, pharmacokinetic and toxicokinetic

studies are the major area that should continue to thrive as microdialysis techniques


Microdialysis in Free-Moving Large Animals

The application ofmicrodialysis to large animals is very desirable in view of the

economic advantage by reducing the number of animals required for experiments. This is

also of special importance when experimental subjects are not readily available.

Technically, in view of the relatively smaller probe to tissue ratio in larger animals, the

use of microdialysis technique in large animals presents some important advantages over

that of small animals. For example, a probe with longer membrane length can be

employed for improved recovery; spatial resolution and relative tissue damage were also

less of a concern in larger animals. However, these advantages also come with various

challenges that are unique to the application of microdialysis in large domestic animals.

For example, in addition to the general concern associate with tissue damage, probe

recovery, and temporal resolution in small animals; unlike its use in small laboratory

animals, the entire microdialysis system must be affixed to the experimental subject in a

manner that it is relatively light weighted yet sturdy to accommodate normal movement

and protect from possible physical damage. Furthermore, large animals require much

longer sections of narrow-bore tubing, problems with slow, inconsistent, or blocked

perfusion flow are more likely to occur. To add one more, probes used in large animals

have generally undertaken higher tissue pressure and are relatively more subject to

damage by tissue movement. These concerns require special precautions in designing

microdialysis experiments; therefore, despite limited success in sheep, microdialysis has

not been widely employed in free moving large animals.


Microdialysis is a developing technique that has tremendous advantages for in

vivo studies involving the pharmacokinetics and pharmacodynamics of drugs. It is an

invaluable research tool for pre-clinical investigations, especially when extensive

sampling from central vasculature or CNS is required. The fact that there is minimal or

no net loss of sampled matrix and the ability to sampling from multiple tissue sites

simultaneously enable the use of a single animal for multiple purposes throughout a

study. It also allows the subject to serve as its own control, such that the number of

experimental subjects could be reduced in a cross-over design; this has great impact on

studies with large animals where bigger sample size is usually costly. To our interest, the

simultaneous use of multiple probes to study tissue-dependent drug distribution,

metabolism and toxicity present the most exciting areas for the application of

microdialysis technique. The themes of this dissertation are to study the central stimulant

effects of CA in the horses, by both cellular and behavioral measures and to study their

relationship to CA levels in CNS and vascular blood as determined by free-animal

microdialysis. With these studies, it should become clearer whether CA can cause

performance-related behavioral change and what effective levels are required for

behavioral changes to occur. These results are of significant importance in considering an

allowable levels for CA in athletic horses. It is also our hope that the completion of this

dissertation would encourage more researchers to take advantage of microdialysis

technique to facilitate the study of pharmacology and pharmacokinetics in larger animal




A major goal of clinical pharmacokinetics is to define the relationships) between

the concentration of a drug and time with its therapeutic or toxic effects in a given animal

species. Temporal measurements of drug concentrations in blood (or serum) may be

fitted to various models in order to derive pharmacokinetic parameters that are used to

predict the time course for drug actions. However, for many drugs, the relationship

between pharmacokinetics and pharmacodynamics is complex and has little predictive

value. One possible explanation for the inaccuracy of pharmacokinetic-derived

predictions is the presence of significant differences between drug concentrations in the

systemic circulation versus tissues or organs wherein the drug acts. Many factors may

contribute to drug concentration gradients between tissues and blood, including plasma

protein binding, restricted drug permeation through membrane barriers and tissue-specific

biotransformation (Nolting, et al., 1996). While direct tissue sampling via repetitive

(serial) tissue biopsies or similar experimental approaches afford alternatives to modeling

based on blood sampling data, these methods are expensive, labor intensive and give rise

to concerns over animal welfare as well as the use of excessive numbers of experimental


Microdialysis is an in vivo sampling method that was introduced for continuous

monitoring of neurochemical events in the brains of small laboratory animal species (Bito

et al., 1966). In view of its capacity to provide a direct measure of the free (unbound)

concentration of drugs in extracellular fluid, microdialysis has been used to study the

distribution and metabolism of drugs in vivo (Johansen et al., 1997; Hansen et al., 1999)

and to measure drug binding to serum proteins (Le-Quellec et al., 1994). In theory,

dialysate samples may be collected from any tissue on a continual basis from a single

animal, thereby reducing the required number of experimental subjects. Nevertheless,

practical limitations associated with instrumentation have limited the use of

microdialysis-based investigations in awake unrestrained animals to several small

laboratory animal species (rats, mice). Although microdialysis has been performed in

deeply sedated or anesthetized animals, including both small (Nakashima et al., 1994;

Sarre et al., 1995) and large animal species (Kendrick, 1991; Ingvast-Larsson et al.,

1992; Shaw and Britt, 1995; Blair et al., 1997; Ohtani et al., 1998), concerns over

possible confounding effects by CNS depressants have restricted its widespread use.

In the present report, we describe a semi-automated method for microdialysis-

based monitoring of drug concentrations in blood and skeletal (splenius) muscle of

unrestrained adult horses. Caffeine was selected as the prototype agent for this

investigation in view of the efficient recovery possible for low molecular weight

compounds that exhibit low protein binding. Simultaneous measurements of two CA

metabolites, TP and TB, were carried out in order to demonstrate the utility of this

approach for studies of drug metabolism. Results from this investigation demonstrate the

practicality of using in vivo microdialysis in lieu of direct blood sampling in order to

monitor circulating drug levels in horses and, furthermore, underscore the advantages of

this method for evaluating drug kinetics in muscle and other soft tissues.

Materials and Methods

Experimental Subjects

One female Thoroughbred (510 kg; single trial) and one female Quarter Horse

(450 kg; two trials) were used for this study. The horses were housed in a fenced pasture

prior to use in this investigation. On the morning of each experiment, a horse was

transferred to a lighted temperature-controlled interior stall (12 by 12 ft.) and given free

access to hay and water. Except for a short period during which microdialysis probes

were implanted, horses were kept in this stall for the duration of the experiment. The

experimental protocols in this investigation that involved horses were reviewed and

approved by the University of Florida Institutional Animal Care and Use Committee prior

to the performance of this study.

Microdialysis Apparatus and Probe Implantation

The semi-automated microdialysis system was assembled entirely from

commercially available equipment and supplies. All components were secured in a clear

plastic box (44 x 29 x 17 cm, 1 x w x h dimensions) that was bolted to a wooden base and

affixed to an exercise saddle on the horse's back (Fig. 2-1). Linear-type microdialysis

probes (LM-10; Bioanalytical Systems, Inc., Lafayette, IN, USA) with a 0.25 mm

external diameter and 10mm membrane length (6 K Da molecular weight cutoff) were

used in all subjects. Following implantation in the jugular vein or splenius muscle (see

below), each probe was connected to a gas-tight microliter syringe (2.5 mL) with small-

bore polyethylene tubing (0.12 mm inner diameter (I. D.); CMA Microdialysis, Inc.,

Stockholm, Sweden). Probes were perfused independently with sterile 0.9% (w/v)

sodium chloride solution (normal saline) with the aid of a small battery-powered infusion

pump (CMA-107, CMA Microdialysis, Inc., Stockholm, Sweden). Following the initial

washout period (see below), probes were perfused at a flow rate of 2 gL/min throughout

the experiment. A maximum length of 1.5meters of pre-sterilized polyethylene tubing

was used to connect the lines of each probe to an infusion pump and microfraction

collector, respectively. The delay time associated with dead volume of the tubing was

determined and accounted for in all data calculations. Dialysate fractions were collected

at preset intervals (5 to 90 minutes) in glass microvials (300 pL capacity) with a

programmable dual-channel microfraction collector (CMA-142, CMA Microdialysis,

Inc., Stockholm, Sweden). The collector was powered by a sealed rechargeable 12-volt

lead battery that was mounted within the instrument box. A water-soaked sponge was

placed in the box in order to minimize evaporative loss ofdialysate fractions from sample

vials. All items (excluding probes) were affixed firmly to a wooden base in the clear

plastic box and covered with a tight fitting lid. The total weight of the saddle, wooden

base and box with all equipment was approximately 23 lb (10.4 kg).

In order to implant microdialysis probes in the jugular vein and splenius muscle,

horses were restrained in a nearby stock and injected with heparin sodium (40 units/kg,

iv) in order to reduce the formation of blood clots on the membrane of the dialysis probe.

Small regions of skin overlaying the left jugular vein and left splenius muscle were

shaved and surgically prepared with an antiseptic solution (Exidine-4; Baxter, Deerfield,

IL, USA). A sterile 14-gauge Teflon catheter (5" length; Abbocath-T, Abbott Ireland,

Ireland) was inserted into the jugular vein and the stainless steel stylet needle was

removed. A microdialysis probe was folded gently in half and passed through the Teflon

catheter so that the tip of the probe extended approximately 6 cm beyond the end of the

catheter. The catheter was affixed to the adjacent skin with adhesive and the hub was

sealed with bone wax. Inlet and outlet lines from the probe were connected as described

above and the probe was flushed with sterile saline at a flow rate of 5 tL/min for a period

of 30 minutes. The procedure for insertion of a microdialysis probe in splenius muscle

was slightly modified from that described above. The stainless steel stylet was removed

from a 14-gauge catheter (Abbocath-T) and bent to form a gentle arc. The stylet was

passed quickly through the splenius muscle in a cranial-caudal direction and parallel to

the presumed direction ofmyofibrils within the splenius muscle. The stylet was inserted

to a maximum depth of 2.5 to 3 cm and the distance between entry and exit points

through the skin was approximately 8 cm. Immediately following needle insertion, a

LM-10 microdialysis probe (unfolded) was passed through the needle and positioned so

that the membrane was centered between the entry and exit points through the skin. The

probe was held in place and the needle was withdrawn slowly, thereby placing the probe

directly within the splenius muscle. Inlet and outlet lines were connected and the probe

was flushed as described above. Following implantation, microdialysis probes were

perfused with sterile saline for a minimum of 1 hour in order to ensure equilibration prior

to CA administration.

Drug Administration and Sample Collection

Immediately before each experiment, anhydrous CA (Sigma Chemical Co., St.

Louis, MO, USA) was dissolved at a final concentration of 60 mg/mL in sterile saline

that contained 48 mg/mL sodium benzoate (Sigma Chemical Co., St. Louis, MO, USA).

CA solution was administered by rapid injection (approximately 30 sec) into the jugular

vein at a dose rate of 3mg/kg. Samples of mixed venous blood (7 mL) were collected in

heparinized vacutainer tubes from the contralateral jugular vein immediately before and

5, 10, 15, 20, 25, 30, 35, 40, 45, 50, 55 minutes and 1, 1.5, 3, 6, 12, 18, 24, 30, 36,42 and

48 hours following CA administration. Blood samples were stored on ice before plasma

was separated and stored frozen at -20 C. Dialysate samples were collected from

splenius muscle and jugular vein according to the following schedule: (1) 5-minute

collection intervals during the initial 50-minute period following CA administration; (2)

30-minute collection intervals during the next 5-hour period; and (3) 90-minute collection

intervals throughout the remainder of the 48-hour collection period. Dialysate samples

were collected in pre-weighed vials in order to detect and correct for variations in the

volume of individual fractions. Microvials containing dialysate samples were covered

tightly with Parafilm and stored at 4 C prior to analysis on the following day.

Drug Analyses

Concentrations of CA, TP and TB in blood samples and tissue dialysates were

determined by high performance liquid chromatography with ultraviolet detection

(HPLC-UV) on a System Gold HPLC system equipped with a Model 166 programmable

detector (System Gold, Beckman, California, USA) and a CMA/200 refrigerated

microsampler (Carnegie-Medicine, Inc., Stockholm, Sweden). CA, TP and TB were

separated (see Fig. 2-2) on a Lichrosphere RP C-18 column (150 x 4.6 mm; 5 gm particle

size; Supelco, Inc., St. Louis, Mo, USA) with mobile phase containing 10 mM NaH2PO4

(pH 4.8), acetonitrile and tetrahydrofuran (90:7:3 v/v) at ambient temperature and 0.8

mL/min flow rate. Samples were injected using a fixed-volume (19.6 pL) loop and MX

were detected at 273 rnm. Drug concentrations in samples were extrapolated from linear

standard curves for each compound (0.005 to 5 p.g/mL). The limits of quantitation for

TB, TP and CA were 5, 10 and 20 ng/mL, respectively.

Plasma samples were subjected to a solid phase extraction (SPE) procedure prior

to drug analysis. The SPE columns (pre-packed C 18 Bond Elute columns, Varian Co.,

Harbor City, CA, USA) were pre-conditioned with 4 mL aliquots of methanol and 10

mM NaH2PO4 (pH = 6.0) prior to the application of buffered horse plasma samples

(1 mL plasma plus 3 mL of 10 mM NaH2PO4; pH = 4.8). Columns were washed with

20 mL of distilled water and dried under vacuum prior to elution with methanol (4 mL).

Column eluates containing drugs were dried in a speed vacuum concentrator and

reconstituted in 1 mL of mobile phase prior to analysis by HPLC. Using this SPE

method, recoveries of drug standards (1 gg /mL) in normal horse plasma were

determined to be 101 3.2 (CA), 93 2.1 (TP) and 98 1.2% (TB). Microdialysate

samples were analyzed without any preparation. However, samples collected at 5-minute

intervals were diluted with normal saline (25 uL) to provide an adequate volume for full-

loop injections.

Pharmacokinetic Analyses

Pharmacokinetic parameters were determined for CA in blood and splenius

muscle for each horse using non-compartmental analysis, which only assumes first-order

kinetics and requires determination of the area under the curve (AUC) from dose time to

infinity, area under the moment curve (AUMC) and the excretion rate. The following

first-order equation:

Ct = C e -Xt + C2 e -2t

was fit to the data from blood (total plasma and jugular vein dialysates) to determine

these values whereas the following first-order equation:

- l~ -i -2et / i ~ \ -X3.t
Ct- = C1 e -t + C2 e -2t (Ci +C2) e 3t

which allows for accumulation as well as disappearance of drug, was fit to the data from

muscle. Ct is the drug concentration at time t; C1, C2, Xi, X2 and X3 are the model values

that are fitted to the data. The weighted non-linear best fit for each mathematical model

was performed with a computerized algorithm based upon minimization of the sum of the

squared deviations (Caceci and Cacheris, 1984). Pharmacokinetic parameters were

calculated based on non-compartmental kinetics (Gibaldi and Perrier, 1982) according to

the following relationships:

AUCjugular = C1/kI + C2/-2

AUCmuscle = C/AI + C2/2 (CI + C2)/3

AUMCjugular = C1/0.12 + C2/022

AUMCmuscle = C1/W 2 + C2/22 (Ci + C2)/3

Kel = 2

t/, = ln(2)/A2


Vdss = dose/AUMC/AUC2

Clearance = dose/AUC

Data Analyses and Statistics

Data were evaluated for statistical differences in several ways. The Student's t

test was used to test for differences between different blood sampling methods (direct

blood sampling vs. microdialysis) and the level of agreement between methods was

evaluated further by the method of Bland and Altman (1986). Results from microdialysis

measurements in the jugular vein and splenius muscle were compared and tested for

differences by a mixed model analysis of variance (ANOVA) with repeated measure over

time. A P level less than 0.05 was accepted as being statistically significant.


In Vitro Calibration of Microdialysis Probes

Preliminary in vitro studies were carried out with the flexible linear-style

microdialysis probes (LM- 10) in order to evaluate the recovery characteristics for CA, TP

and TB under a variety of experimental conditions. Tests revealed that the fractional

(percent) recoveries of CA, TP and TB were identical for all three compounds and that

the performance of the probes was unaffected by changes in the environmental (solution)

temperature (22 vs. 37C) or the rate at which the drug solution was mixed. Although

these results were expected, it was important nevertheless to confirm these expectations

in order to ensure that changes in the experimental conditions would not impact the

results for studies in vivo with horses.

The in vitro recoveries of CA, TP and TB were determined at different

concentrations in several media at three representative flow rates for the LM-10

microdialysis probe. Each compound exhibited similar properties and behaved in the

predicted manner insofar as recovery remained constant over a wide range of drug

concentrations (0.02 to 5 gg/mL, Table 2-1). As expected, the recovery of each

compound was found to be inversely related to the probe flow rate with average recovery

values (n = 3) in normal saline of61 1%, 84 1% and 98 + 1% at flow rates of 2, 1 and

0.5 pL/min, respectively (Table 2-1). In addition, studies confirmed that drug passage

through the dialysis membrane exhibited bi-directional symmetry, insofar as the fraction

of drug recovered from an isotonic medium (recovery mode) was equivalent to the

fraction of drug that diffused from the perfusion fluid into the external drug-free medium

(delivery mode) (Table 2-2). This result was significant insofar as it provided a rational

basis for using the delivery method as way to measure CA recovery in vivo for

microdialysis probes placed within the jugular vein or splenius muscle of horses.

Finally, a brief study was carried out in order to delineate the extent of methylxanthine

drug binding to equine serum proteins. CA, TP and TB were dissolved in fresh horse

serum at a final concentration of 5 utg/mL and incubated (12 hours at 4C) prior to the

estimation of free drug concentrations (CF) by microdialysis. The percent of drug bound

to protein (B) at 37C was calculated from the relationship B = [ 1 CF/CT ] X 100%,

where CT is the total concentration of drug in serum (5 jtg/mL) and CF is the drug

concentration in dialysate divided by fractional drug recovery. Under these conditions,

protein binding was determined to be relatively low for CA (12 1.1%) and TB (14

0.3%), but slightly higher for TP (34 1.4%).

Comparison of CA Concentrations in Venous Blood by Venipuncture and In Vivo

Following intravenous CA administration (3 mg/kg), samples of venous blood

were drawn by venipuncture at pre-determined intervals and compared with time-

matched microdialysate samples from the contralateral jugular vein. As shown in

Fig. 2-3A (left panel), the highest CA concentration of 5400 400 ng/mL in blood was

detected at the earliest blood sampling time (5 minutes). Subsequently, CA concentration

in venous blood decreased at a pseudo first-order rate to a concentration of 380 40

ng/mL by the end of the 48-hour sampling period (see inset). Non-compartmental

pharmacokinetic analysis revealed an apparent t /, of 14.4 0.6 hours for CA with a

mean residence time (MRT) of 20.6 0.8 hours in blood (Table 2-3).

When compared with results obtained by direct venipuncture, it appears that in

vivo microdialysis provided a strikingly similar profile for CA pharmacokinetics. As

shown in Fig. 2-3A (right panel), in vivo microdialysis revealed a nearly identical peak

CA concentration in blood (5400 + 600 ng/mL) with a linear decline throughout the 48-

hour sampling period. Analyses based upon a non-compartmental model revealed no

significant differences between any of the pharmacokinetic parameters for CA in blood as

determined by venipuncture or in vivo microdialysis (Table 2-3). However, in order to

test for possible significant differences between these sampling methods, data were

compared using a statistical approach that is designed to evaluate specifically the level of

agreement between alternative methods for performing a clinical measure (Bland and

Altman, 1986). As shown in Fig. 2-4A, a plot of average corrected CA concentrations in

dialysate samples versus time-matched venipuncture samples revealed the close

proximity of all data points to the predicted line of equality (slope equal to 1).

Correlation analysis affirmed the close agreement between these sampling methods and,

thereby, provided a preliminary basis for rejection of the null hypothesis that drug levels

obtained by these methods are not linearly related. However, despite the strong

correlation (R2 = 0.99) and regression slope (0.99) that approximates closely the

predicted value, additional evidence is needed in order to conclude that measurements

obtained with these methods are interchangeable. In order to reach such a conclusion, it

is necessary to evaluate statistically any observed differences among the experimental

measures through the construction and use of a 'difference' plot (Bland and Altman,

1986). When applied to the present study, arithmetic differences between CA

concentrations in time-matched samples obtained by venipuncture and microdialysis are

plotted against the overall mean values for CA concentrations detected by both methods.

Inspection of the difference plot for CA concentrations in blood (Fig. 2-4B) provides

strong evidence that in vivo microdialysis and venipuncture yield equivalent results for

drug measurements in blood. Support for this view is based on several factors, including

a poor correlation between difference values and the corresponding mean CA

concentrations in blood as well as the absence of any significant inter-method bias. Our

conclusion that bias between microdialysis and venipuncture methods is insignificant is

based upon the small value for the mean population difference (27 ng/mL) and the

equivalent distribution of difference values above or below this point (Fig. 2-4B).

Finally, insofar as inter-method differences in blood concentrations of CA exhibit a

normal (Gaussian) distribution, it would be expected that 95% of all difference values

should fall within two standard deviations of the mean population difference (i.e., 27 + 2

x 123 ng/mL). As shown in Fig. 2-4B, all difference values fall within the predicted


Comparison of CA Concentrations in Venous Blood and Splenius Muscle

In conjunction with these measurements of drug concentrations in blood, CA and

CA metabolite concentrations were monitored simultaneously in splenius muscle by in

vivo microdialysis. As shown in Fig. 2-5A, CA achieved a peak concentration in muscle

(5800 200 ng/mL) that was not significantly different from the peak concentration

measured in venous blood. However, timing of the peak CA concentration in muscle was

delayed by 25 to 30 minutes (sixth fraction) relative to the time of the peak blood CA

concentration. Noncompartmental pharmacokinetic analysis of CA concentrations in

muscle revealed moderate albeit statistically significant reductions in the apparent ti2

(11.8 0.7 hours) and MRT (16.6 0.9 hours) for CA relative to blood (Table 2-3).

Nevertheless, in spite of the delayed distribution and faster removal of CA from muscle

compared to the vascular compartment, CA levels were significantly higher in muscle for

a period of about 4 hours following drug injection. As shown in Fig. 2-6, CA levels in

muscle and blood dialysates are highly correlated (R2 = 0.98), although the slope of the

regression line is substantially greater than unity (m = 1.24). While the latter result is

consistent with a preferential accumulation of drug in muscle, a temporal plot of the

difference in CA concentrations between muscle and blood provides direct evidence for

this difference. As shown in Fig. 2-6B, CA concentrations are significantly elevated

(p < 0.05) in splenius muscle relative to blood throughout the initial 4-hour period

following CA administration. However, aside from these early differences, CA

concentrations in blood and muscle do not differ throughout the remainder of the 48-hour

sampling period.

CA Metabolites in Mixed Venous Blood and Splenius Muscle

In addition to measurements of CA concentrations, in vivo microdialysis was used

to monitor levels of two demethylated CA metabolites in blood and muscle. In

venipuncture samples, TP and TB were detected at 30 and 60 minutes, respectively,

following CA injection. In contrast, these CA metabolites were not detectable in jugular

vein dialysates until 1 hour (TP) and 2 hours (TB) owing to the diminished recovery of

drugs by microdialysis-based sampling. As shown in Fig. 2-3, the levels of TP and TB in

venipuncture and dialysate samples exhibited a gradual rise to peak concentrations of 380

and 100 ng/mL, respectively, approximately one day after CA injection. Although the

concentrations of both metabolites appeared to decrease modestly during the subsequent

24-hour monitoring period, the concentrations of both compounds remained within 20%

of their respective peak concentrations when sampling was halted at 48 hour. Dialysates

from splenius muscle revealed similar profiles for both TP and TB although the

appearance of TB in muscle was delayed by 1 hour relative to jugular vein dialysates.

The timing of the peak concentrations of TP (300 60 ng/mL) and TB (150 50 ng/mL)

in muscle occurred at 18 and 20 hours, respectively.


This report describes and validates a semi-automated microdialysis technique for

continuous sampling of drugs in vivo from mixed venous blood and soft tissue (skeletal

muscle) of unrestrained adult horses. The capacity to simultaneously measure levels of

drugs and drug metabolites by in vivo microdialysis in multiple tissues affords

investigators with a powerful and less invasive approach for conducting pharmacokinetic

studies in horses and possibly other domestic large animal species. In the present study,

horses were injected with a behaviorally-active dose (3 mg/kg, iv) of the psychomotor

stimulant drug CA and drug concentrations in blood were measured by intermittent

venipuncture and continuous in vivo microdialysis. Rigorous statistical comparisons of

raw data and fitted pharmacokinetic parameters demonstrate that experimental data

obtained with either sampling technique are indistinguishable (Figs. 2-3 and 2-4; Table 2-

3) and agree closely with results published previously for horses (Greene et al., 1983;

Aramaki et al., 1991, 1995,1996; Schumacher et al., 1994; Peck et al., 1997). In view of

our findings, we conclude that the semi-automated microdialysis procedure described

here represents a suitable method for the performance ofpharmacokinetic studies in

domestic large animal species. However, microdialysis-based measurements offer a

distinct advantage over traditional venipuncture techniques insofar as interactions with

the experimental subject are greatly reduced and multiple tissues may be sampled

simultaneously. These advantages could be truly significant for studies designed to

elucidate subtle behavioral effects by drugs or related investigations wherein interactions

between the experimenter and experimental subjects must be minimized.

The results of this investigation indicate that measurements of systemic drug

concentrations by continuous in vivo microdialysis and intermittent venipuncture yield

statistically indistinguishable results in horses. However, it should be noted that

significant differences are likely to have been observed if the study were repeated with a

drug that is highly bound to plasma proteins. In such a case, the free drug concentration

in microdialysate samples would be expected to differ substantially from the total drug

concentration in venipuncture samples owing to the inability of bound drug molecules to

cross the dialysis membrane (Wright et al., 1996; Bailey 1998). Therefore, the close

agreement between total and free CA concentrations in the present study (Figs. 2-3 and

2-4) indicates that CA is not highly bound to equine plasma proteins. This prediction was

confirmed through direct studies in vitro, which estimate the protein bound fraction of

CA at 12% in horse blood.

In addition to our comparison of methods for measuring systemic drug levels, in

vivo microdialysis was tested for its suitability to carry out similar measures in equine

skeletal muscle. Following several preliminary trials, the lateral splenius muscle was

chosen for all microdialysis-based studies. Several factors supported selection of the

splenius muscle, including its close proximity (horizontal and vertical distances) to the

instrument box, the parallel orientation of myofibrils throughout the full thickness of the

muscle as well as the superficial location of the muscle, which facilitates reliable

straightforward insertion of the probe. Preliminary trials revealed that many technical

problems (reduced dialysate flow, probe breaks, etc.) could be avoided entirely if probes

were inserted parallel rather than orthogonal to the orientation of the muscle myofibrils.

While the reason for this difference is not entirely clear, we assume that probes oriented

parallel to the myofibril long axis experience less physical stress associated with changes

in muscle length.

As shown in Fig. 2-5, CA concentrations rise quickly in muscle following

intravenous drug administration, whereas muscle concentrations of TP and TB exhibit

significant delays. While visual inspection of the concentration-time plots for CA in

muscle (Fig. 2-5A) and blood (Fig. 2-3A) suggest nearly superimposable profiles,

detailed comparisons uncover several significant differences. As shown in Table 2-3, the

apparent t / and mean residence time for CA are reduced by greater than 20% in splenius

muscle relative to blood. In addition, CA concentrations in muscle are significantly

elevated above blood concentrations for a period of 4 hours following CA injection (Fig.

2-6B). In agreement with the latter observation, correlation analysis of CA levels in

muscle and blood reveals a regression line having a slope (1.24) significantly greater than

unity (Fig. 2-6A). While the basis for these differences is unknown, similar disparities

have been reported previously for CA distribution in rats (Stahle et al., 1991 b) and

humans (Stahle et al., 1991c). In considering potential causes for these differences, it is

important to evaluate both technical and biological factors that could contribute to tissue-

specific pharmacokinetics of CA in horses. With regard to possible technical causes,

site-related differences in drug recovery should always be considered (Stahle, 1991 a). In

this study, CA recovery was measured in vivo using the delivery method (see above) and

estimates of fractional recovery were used to derive tissue drug concentration from drug

concentration in dialysate. While this approach has some inherent inaccuracy, any errors

associated with recovery estimation should remain constant throughout the sampling

period. However, insofar as differences between CA levels in muscle and blood exhibit a

reproducible temporal pattern (Fig. 2-6B), other possible sources of the difference must

be considered. Local changes secondary to implantation of the probe could promote

elevated drug concentrations in muscle ECF. For example, guide needle insertion and

placement of the probe within muscle could cause local irritation with an associated

inflammatory reaction. In response to local inflammation, regional changes in capillary

permeability or the muscle microenvironment could take place leading to altered drug

delivery or retention of CA via a secondary (e.g., ion -trapping) mechanism.

Alternatively, it is possible that muscle accumulates CA by active transport in a manner

similar to described for the central nervous system (McCall et al., 1982; Nehlig et al.,

1992). Nevertheless, while the basis for the brief accumulation of CA in equine muscle

remains unclear, our results underscore the importance of monitoring drug levels directly

at potential sites of tissue action. In view of the ability of CA to directly influence

skeletal muscle contraction (Macintosh et al., 1981), measurement of CA concentrations

in muscle tissue may be necessary to provide better insight into the overall effects of this

drug on performance and endurance.

Several reports have addressed the advantages and limitations of microdialysis-

based approaches for pharmacokinetic studies (Ungerstedt, 1991; Elmquist et al., 1997;

Johansen et al., 1997; Hansen et al., 1999), although those discussions have been limited

to small animals. The use of this technique in large domestic animals presents additional

problems. Unlike its use in small laboratory animals, the entire microdialysis system

(probes, connecting tubing, pumps and fraction collector) must be affixed to the

experimental subject in a manner that accommodates normal movement. In view of this

constraint, the system should be relatively small and lightweight, yet protected from

possible physical damage. The self-contained system described here (Fig. 2-1) satisfies

these requirements and is readily assembled (with minor modifications) from equipment

and products that are available commercially. However, in using this technique, one

must consider the location of sampling sites relative to the microperfusion pumps and

fraction collector. Since large animals require much longer sections of narrow-bore

tubing to connect dialysis probes with pumps and collectors, problems with fluid flow

rates are more likely to occur. Excessive tubing lengths create increased resistance to

flow whereas differences in the vertical position of probes relative to the instrument box

may increase the hydrostatic pressure. In our hands, the battery-powered syringe pumps

were able to maintain flow rates within 10% of the nominal rate for tubing lengths up to

1.5 meters. However, flow rates dropped precipitously when microdialysis studies were

attempted with longer runs of tubing or when probes were located substantially below

(greater than 30 cm vertical distance) the instrument box. For the present investigation,

probes were implanted in sites (jugular vein and splenius muscle) that were located

nearly horizontal to the perfusion pumps and collector in order to avoid these problems.

In conclusion, we have developed and validated a reliable in vivo sampling

technique for horses based upon tissue microdialysis. The procedure allows for

continuous measurements of free drug levels in blood and soft tissue for a period of

several days without significant problems. Studies with CA have confirmed the accuracy

of this sampling method in blood and have revealed significant time-dependent

differences between the free concentration and pharmacokinetics of CA in blood versus

skeletal muscle. This approach should be readily adapted for use in other large domestic

animal species and could improve our understanding of clinical pharmacokinetics as well

as the capacity to estimate drug residues in specific tissues of interest.

Table 2-1. In vitro recovery of CA, TP and TB in normal saline. The recovery of probe
(CMA 20) was tested at combinations of 3 perfusion rates and 3 levels of MX. Numbers
shown are percentage recovery (mean SEM) for 3 experiments.

Concentration Low Med High
Levels 20-50 200-500 2-5
(ng/mL) (ng/mL) (tg/mL)
Perfusion rate TB 58 4 58 4 58 4
2 TP 62 4 68 4 65 4
pL/min CA 57 3 63 5 62 5

TB 82 1 84 4 853
1 TP 86 3 84 1 86 2
gL/min CA 83 3 83 4 85 3
TB 98 1 98 0 98 1
0.5 TP 98 2 98 1 98 1
giL/min CA 98 2 98 2 98 1

Table 2-2. Comparison of in vitro delivery and in vitro recovery of CA, TP and TB
(1 tg/mL) in normal saline (NS) and normal horse serum (NHS). Experiments were
performed at 1 tL/min perfusion rate.

MX Recovery Delivery Delivery Recovery
in NHS (%) in NHS(%) in NS (%) in NS (%)
TB 83 84 82 84
TP 71 71 85 85
CA 76 79 82 84

Table 2-3. Derived pharmacokinetic parameters for CA in venous blood and splenius
muscle. Values for three independent trials are presented as well as mean ( SEM) in
horses administered intravenous CA (3 mg/kg).

Horse 1 Horse 2 Horse 3 Mean SEM

(A) Direct blood drawing
t /,(h) 14.3 15.3 13.6 14.4 0.6
Vd ss (L/kg) 0.78 0.91 0.65 0.78 0.09
Cl (mL/min/kg) 0.64 0.70 0.55 0.63 0.05
AUC (gg*h/mL) 79.6 71.7 89.8 80.4 6.4
MRT (h) 20.4 21.8 19.5 20.6 0.8

(B) Microdialysis in Jugular vein
t ,(h) 16.3 15.3 14.3 15.3 0.7
Vd ss (L/kg) 0.85 0.89 0.71 0.82 0.07
Cl (mL/min/kg) 0.67 0.69 0.58 0.65 0.04
AUC (ug*h/mL) 76.4 72.1 85.9 78.1 5.0
MRT(h) 21.3 21.5 20.4 21.1 0.4

(C) Microdialysis in Splenius muscle
t /,(h) 10.7 12.4 12.3 11.8 0.7*
Vd ss (L/kg) 0.69 0.63 0.50 0.61 0.07
Cl (mL/min/kg) 0.75 0.59 0.50 0.61 0.09
AUC (gg*h/mL) 67.6 85.1 100.3 84.3 11.6
MRT (h) 15.3 17.9 16.8 16.6 0.92*

* Significantly different from values of the same parameter in panel (A) and (B)


Instrument Box

, ....... ... . ....
I7 / \''A

"" "- ,' 1 TOP VIEW
/ ;S ,77 ,

/ ** '*: J I i-m B'J
Muscle Jugular 1- i ... L oI,
Probe Probe _, ,-

Fig. 2-1. Schematic representation of instrument box and implantation sites for
microdialysis probes in horses. A semi-automated microdialysis system was assembled
by placing a microfraction collector, a 12 Volt DC battery, and two battery-powered
mini-pumps in a clear plastic box with a removable lid. The box was bolted to a wood
base and affixed to an exercise saddle on the horse's back. A water-soaked sponge was
placed in the box in order to minimize evaporative loss of dialysate samples. The total
weight of the saddle, wooden base and box with all equipment was approximately 23 lb.
A maximum length of 2 meter of fine-bore (12.5 UL/m) tubing was braided through the
mane to connect pump/collector to each probe. Probes were implanted in left jugular vein
and splenius muscle.









0 1 2 3 4 5 6

0 1 2 3 4 5 6

Time (minute)

Fig. 2-2. Sample chromatograms for HPLC separation and UV absorption of
methylxanthine drugs in equine blood. Analyses were conducted using samples of drug-
free (control) plasma following solid phase extraction (panel A) or microdialysate
samples from whole blood containing 1 pg /mL of each drug (panel B). The probe
perfusion rate was 0.5 pUL/min. Average retention times for CA, TP and TB were 5.4, 4.2
and 3.1 minutes, respectively.

Direct Blood Sampling Microdialysis

(A) Caffeine -
6000 |i

.4000 8
102C 0
S0 12 24 36 48 0 12 24 36 48
2000 TM (hr) T IME hri

0 12 24 36 48 0 12 24 36 48
-S (B) Theophylline

: 450
'* / ^' *~-- .-.
2 300 i
1, 5:- ,__ il --_-
= 150
0 0
S 0 12 24 36 48 0 12 24 36 48

200 (C) Theobromine


50 2

0 12 24 36 48 0 12 24 36 48

Time (h)

Fig. 2-3. Concentration-time profiles for MX in systemic circulation following CA
administration (3 mg/kg, i.v.). Concentrations (mean SEM, n = 3) of CA, TP and TB in
mixed venous (jugular vein) blood were determined by HPLC-UV analysis of samples
obtained by venipuncture (left panels) or in vivo microdialysis (right panels). For
microdialysis experiments, bar widths indicate the duration of individual sampling
intervals (5, 30 and 90 minutes). Inset figures depict semi-logarithmic transformations of
averaged data for CA.




0 2000 4000

Dialysate [CA] (ng/ml)



- a


-400 -





Average [CA] (ng/mL)

Fig.2-4. Comparison of jugular vein CA concentrations measured by direct sampling and
in vivo microdialysis. All data are derived from results shown in Fig. 2-3A. (A) Scatter
plot of average (n = 3) jugular vein CA concentrations at time-matched points as
determined by the two sampling techniques. The regression line and line of equality are
plotted as solid and dashed lines, respectively. The correlation coefficient (R2) and slope
for the regression line were both 0.99. (B) Difference plot ofjugular vein CA
concentrations for both sampling techniques. The arithmetic difference between average
CA concentrations in matched samples (A CA) were calculated and plotted as described
by Bland and Altman (1986). Dashed lines and shading depict the overall mean A CA
(27 ng/mL) two standard deviations.

Mean + 2SD

S Mean

-- Mean ----2SD
Mean 2SD

o6000 Caff'e i e ., ,-

4O6 103
4000 Z

2000 M. -10'0 1, 24 36 48
h-I M TIME(hr)
0 0 12 24 36 48
(B) Theophylline

300 T



S0 12 24 36 48

225 (C) Theobromine

150 T

0 12-'--

0 12 24 36 48
Time (h)

Fig. 2-5. Concentration-time profiles for MX in splenius muscle as determined by in vivo
microdialysis. Data represent average drug concentrations (mean SEM) from three
independent experiments. Inset to panel A depicts semi-logarithmic transformation of
averaged data for CA. Fitted parameters from pharmacokinetic analysis of these data are
summarized in Table 2-3.


6000 o- 7--
-.' Slope = 1.24 .'

= 4000 R2= O.98


0 2000 4000 6000
Jugular vein [CA] (ng/mL)


,. 1500


2 0
-500 J..
0 10 20 30 40 50
Time (h)

Fig. 2-6. Temporal comparison of CA concentration in splenius muscle versus jugular
vein. Average (n = 3) CA concentrations in time-matched dialysate samples are plotted
for the entire sampling period. (A) Scatter plot and linear regression of data obtained in
muscle and blood. Linear regression (solid line) revealed a correlation coefficient of 0.98
and a slope of 1.24. Dashed line represents the line of equality. (B) A temporal plot of
differences between CA concentrations (A CA) in splenius muscle and jugular vein.
Open circles indicate times (2.5 to 245 minutes except for 7.5 and 47.5 minutes) at which
the CA concentration was significantly higher (p < 0.05) in muscle dialysates.



Methylxanthines, especially CA and its metabolites TP, TB and XN are among

the most widely used CNS stimulants. Three main hypotheses have been formulated

concerning possible biochemical mechanisms of action of CA and its metabolites at the

cellular level, which are: intracellular mobilization of calcium, inhibition of PDE activity,

and antagonism of adenosine receptors (Nehlig et al., 1992). While the actions on

calcium and PDE cannot be excluded from certain behavioral (Daly, 1993) and

cardiovascular effects of CA (Howell, 1993a), animal studies have shown that both PDE

inhibition and calcium mobilization occur at CA levels much higher than those required

for adenosine receptor blockade (mM vs. low nM) (Guthrie et al., 1967; Weiss and Hait,

1977; Cardinali et al., 1980; Wachtel et al., 1982), which is regularly achieved in plasma

at typical human coffee consumption (Nehlig et al., 1992; Fredholm, 1995) and at

therapeutic doses of CA in animals (Greene et al., 1983; Daly, 1993; Peck et al., 1997).

Therefore, studies during the past decade have suggested that binding to adenosine

receptors and antagonism of the actions of agonists at these receptors is the most

significant mechanism for the major effects of CA including CNS stimulation in most


Pharmacological and molecular cloning studies have revealed the existence of at

least four distinct adenosine receptors, Al, A2a, A2b and A3 (Fredholm et al., 1994).

While adenosine exhibits very low affinity (> 10 tM) for A2b and A3 classes of

receptors, extracellular adenosine concentrations in most tissues under normal

physiological conditions are high enough to activate Al and A2a receptors. Thus, Al and

A2a receptors have been the major targets for study of effects of CA and related NMX

(Fredholm, 1995). Both Al and A2a receptors are G protein coupled receptors that elicit

opposite control over membrane-bound adenylate cyclase (Van Calker et al., 1979; Bruns

et al., 1980; Daly, 1982). Via actions on Al receptors, adenosine decreases neuronal

firing (Kostopoulos et al., 1977; Phillis et al., 1983) and inhibits the release of excitatory

neurotransmitters (Fredholm et al., 1980; Hollins et al., 1980) while via actions on A2a

receptors, adenosine can decrease the affinity of dopamine binding to dopamine D2

receptors (Ferre et al., 1991; 1992) and block post-synaptic D2 receptor mediated actions

(Jin et aL, 1993). Therefore, by binding to Al and A2a receptors and antagonizing

adenosine effects, CA and related MX would be expected to promote central stimulation

and increase motor related activities (Daly et al., 1981). The anatomical distribution of

Al and A2a receptors in mouse brain is consistent with the proposed actions insofar as

receptor autoradiography has demonstrated that while Al receptors are widely distributed

in the brain with highest levels in cortex, hippocampus and cerebellum, whereas A2a

receptors are concentrated in striatum and dopamine-rich regions of the brain (Fredholm,


The affinity of MX at central Al and/or A2a receptors and their rank order

potencies have been studied in various animal species including rat, mouse, guinea-pig,

cattle and human (Schwabe et al., 1985; Ukena et al., 1986; Daly, 1993; Maemoto et al.,

1997). However, similar studies have not been carried out in the horse despite the fact

that central stimulant effects of CA have been demonstrated in the horse (Greene et al.,

1983) and the potential for non-medical use of CA and its metabolites to improve athletic

performance is high. In order to understand the central stimulant effects of CA in the

horse, characterization of adenosine receptor binding in equine forebrain tissue is

desirable in view of the fact that marked species differences in the affinity (Klotz et al.,

1991; Nonaka et al., 1996), G-protein coupling (Jockers et al., 1994; Nanoff et al., 1995)

and immunological recognition (Nakata, 1993) have been described previously. In

addition, there is a marked pharmacokinetic difference for CA inasmuch as a much

longer t '/ was found in horses (14 hours) (Greene et al., 1983; Peck et al., 1997)

compared to humans (3 hours) and rodents (1 hour) (Fredholm, 1995). Therefore, the aim

of this study was to pharmacologically characterize adenosine Al and A2a binding sites

in horse cortex and striatum using selective radioligands and to evaluate the affinity and

potency of selected MX including CA, TP, TB and XN at equine Al and A2a receptors.

In view of the nature of ligands used in this study (Muller et al., 2000), the

characterization of antagonistic properties of these MX at adenosine receptors was

accomplished by measurement of agonist-stimulated [35S] -guanosine 5'-(y-thio)

triphosphate ([35S]GTPyS) binding to receptor-bound G proteins, which reflects the GDP-

GTP exchange (G protein activation) that was activated upon agonist binding to the

receptors. Binding of antagonists would be expected to lack such an effect. The

stimulation of [35S]GTPyS binding by Al receptor-bound G protein has been

demonstrated in bovine cortex (Lorenzen et al., 1993) and rat brain membranes (Laitinen,

1999; Moore et aL, 2000), but evidence for A2a receptor-mediated [35S]GTPyS

stimulation is still insufficient. The current study represents the first characterization of

[35 ]GTPyS binding assay in equine brain tissues by adenosine receptor agonists and


Material and Methods


8-[dipropyl-2,3-3H (N)]-cyclopentyl-1,3-dipropylxanthine ([3H]DPCPX, 111.6

Ci/mmol) and [35S]GTPyS (1250 Ci/mmol) were purchased from New England Nuclear

(Boston, MA, USA). [2-3H](4-(2-[7-amino-2 (2-furyl)[ 1,2,4]triazolo[2,3-a] [1,3,5]triazin-

5-ylamino]ethyl)phenol) ([3H]ZM241385, 17 Ci/mmol) and non-tritiated ZM241385

were purchased from Tocris Cookson Inc. (Ballwin, MO, USA). Guanosine-5'-O-(3-

thiotriphosphate) (GTPyS) was purchased from Research Biochemical International

(Natick, MA, USA). Adenosine deaminase (ADA) was purchased from Boehringer

Mannheim (Indianapolis, IN, USA). Dimethyl sulfoxide (DMSO), adenosine, bovine

serum albumin (BSA), dithiothreitol, GDP, EDTA, CA, TP, TB, and XN were purchased

from Sigma (St. Louis, MO, USA). Tris, NaOH, NaCI and MgCh2 were purchased from

Fisher Scientific (Pittsburgh, PA, USA). 9-chloro-2-(2-furanyl)-5,6-dihydro-[1,2,4]-

triazolo[1,5]quinazolin-5-imine monomethane sulfonate (CGS15943) was a gift from

Novartis Pharmaceuticals Corporation (Summit, NJ, USA). In view of the poor solubility

in water, CGS 15943 (5 mM) was routinely dissolved in DMSO and diluted 100-fold in

Tris-HCl buffer before each experiment.

Membrane Preparation

Fresh equine brain was obtained from the Anatomic Pathology lab in the College

of Veterinary Medicine, University of Florida. The cerebral cortex and striatum were

dissected and the tissues were homogenized separately with a polytron homogenizer

(setting #4, 2 x 15 sec., Brinkmann PT 10/35, Westbury, NY, USA) in 30 volumes of ice-

cold water. The homogenates were centrifuged at 48,000 x g, 4 C for 15 minutes and the

resulting pellet was resuspended in 30 volumes of ice-cold 50 mM Tris-HCl buffer (pH

7.4) with polytron (setting # 4, 15 sec) and centrifuged at 48,000 x g, 4 C for 15 minutes.

After repeating the wash step with Tris-HCl buffer, the final pellet was resuspended in 5

volumes (original tissue wet weight) of 50 mM Tris-HCl buffer (pH 7.4) at a protein

concentration of approximately 7 mg/mL for cortex and 9 mg/mL for striatum, as

determined by the Lowry method (Lowry et al., 1951) using BSA standards. Aliquots

(1 mL) were stored at -80 C until used for binding assays.

Binding of [3HJDPCPX to Membranes

Saturation binding studies were carried out in assay buffer containing 50 mM

Tris-HCl, pH 7.4, 0.1 unit/mL ADA and 1% DMSO. Fresh thawed frozen aliquots of

membrane preparations were diluted in 50 mM Tris-HCl and added to the assay buffer at

a final amount of 30 to 40 pg protein (cortex) or 40 to 50 ug protein (striatum). The

binding reaction was initiated by adding [3H]DPCPX ( 0.05 to 10 nM final concentration)

in a final volume of 500 tL. After a 15-minute incubation at 25 C, tissue-bound

radioactivity was separated from unbound ligand by vacuum filtration of assay solutions

through GF/B filters (Brandel Inc. Gaithersburg, MD, USA) by using a Brandel cell

harvester (Brandel Inc. Gaithersburg, MD, USA) and washed 3 times with 2 mL of ice-

cold Tris-HCl buffer (50 mM, pH 7.4). Filter disks were transferred to scintillation vials

containing 2 mL of 0.2 N sodium hydroxide (NaOH) and incubated at 37C for 3 hours to

extract filter-bound tritium. Radioactivity was determined by adding 1 mL of 0.5 N HC1

and 12 mL of EcoLume scintillation fluid (ICN Biomedicals Inc., Costa Mesa, CA, USA)

and counted in a LKB scintillation counter (Pharmacia LKB Nuclear Microtomy Inc.,

Gaithersburg, MD, USA). Non-specific binding was determined in the presence ofnon-

xanthine adenosine receptor antagonist CGS 15943 (5 ULM in 1% DMSO) and accounted

for approximately 8% of total binding under these conditions. For competition binding

experiments, membrane suspensions were incubated with a fixed concentration of

[3H]DPCPX (1 nM final concentration) and 9 log-spaced concentrations (0.1 gM to 10

mM final concentration) of competing drugs (CA, TP, TB, XN) in assay buffer (total

volume 500 AL) at 25C for 15 minutes. The experiment was terminated and

radioactivity was determined as described above.

Binding of [3HIZM241385 to Membranes

The standard assay for [3H]ZM241385 saturation binding experiment was

comprised of 50 mM Tris-HCI-NaCl-KCl buffer (50 mM-Ttris-HCl containing 120 mM

NaCI and 5 mM KCI, pH 7.4), 1 unit/mL ADA, 1% DMSO and 300 to 500 ig

membrane protein. The binding reaction was initiated by adding various concentrations

(0.05 to 20 nM) of [3H]ZM241385 in a final volume of 500 4L. The incubation was

carried out for 30 minutes at 25C and was terminated as described for [3HJDPCPX

binding except for using 50 mM Tris-HCl-NaCl-KCI for washing. For competition

binding experiments, 3 nM of [3H]ZM241385 was incubated with 9 log-spaced

concentrations (0.1 RM to 10 mM final concentration) of competing MX in assay

conditions otherwise identical to saturation binding experiment. Non-specific binding

was determined in the presence of 5 uM CGS 15943 (in 1% DMSO). In separate

experiments, in order to verify the selectivity of [3H]DPCPX in striatal membranes,

various concentrations ofZM241385 (0.1 nM to 10 uM final concentration) were used to

compete with [3Hj]DPCPX (1 nM) binding or [3H]ZM241385 (3 nM) binding at equine

cortex and/or striatum membranes. These competition-binding assays were carried out at

conditions identical to respective ligand binding experiments described above.

Binding of [35SJGTPyS to Membranes

Activation of G protein by adenosine receptor agonists and antagonists was

evaluated through [35S]GTPyS binding analysis. [35S]GTPyS binding experiments were

performed by a modification of a previous method (Lorenzen et al., 1993). Receptor-

linked G protein activation was determined by measuring the stimulation of [35S]GTPyS

binding at equine cortical and striatal membranes. The maximum activation of G protein

was assessed by a saturation curve examining the binding of [35S]GTPyS to membrane

preparations at the presence of increasing concentrations of the natural agonist adenosine.

Preceding the experiment, a membrane wash-step was performed (cortex 70 lag protein or

striatum 50 lag protein) by adding 30 volumes of 50 mM Tris-HCl-NaCl-KCl buffer to

the membrane preparations and incubated at 37 C for 20 minutes. The pretreated

membranes were then centrifuged (48,000 x g, 4 C for 15 minutes) and reconstituted to

the original volume and incubated with various concentrations of adenosine (0.2 nM to

15 gM final concentration) in a buffer containing 50 mM Tris-HCI pH 7.4, 80 mM NaCl,

4 mM MgC12, 1 mM EDTA, 1 mM dithiothreitol, 10 uM GDP, 0.5% BSA and 0.2 nM

[35SI]GTPyS at a total volume of 300 LL, for 30 minutes at 25C. Non-specific binding

was defined with 20 gM of GTPyS. The experiment was terminated as described for

[3IH]DPCPX binding. For determination of antagonistic efficacy and potency, competition

binding experiments were carried out by including CGS 15943 and MX in the

abovementioned [35S]GTPyS binding assay in the absence or presence of 4 .M


Data Analysis

In all experiments, binding assays were carried out in duplicate and repeated over

three to four separate experiments. Unless specified otherwise, all results are expressed as

arithmetic means SEM. For [35S]GTPyS experiments, data were plotted as percent of

maximal stimulation level which was designated as the magnitude of agonist (4 pM

adenosine)-stimulated [35S]GTPyS binding relative to unstimulated control. Saturation

binding data were transformed into Scatchard plots. The binding affinity (Kd) and

maximum binding capacity (Bmax) were calculated from computer assisted linear

regression curve fitting (Sigmaplot, Jandel, USA) of bound (B) ligand concentration vs.

the ratio of bound/free (B/F). The resultant straight line had a slope of -1/ Kd and an X-

intercept at Bmax. For competition binding data, each competition binding curve was

analyzed by weighted nonlinear regression curve fitting (Sigmaplot, Jandel, USA) to the

logistic equation to determine IC50 (the concentration that inhibits 50% of ligand binding

in each affinity state). IC50 values were converted to Ki (inhibition constant) values using

the Cheng-Prusoff equation: Ki = IC50 / (1 + [Ligand]/ Kd) where Kd values were

obtained from saturation binding experiments (Cheng and Prusoff, 1973). Values of Kd

and Ki are reported as geometric means in view of the log-normal distribution of these



[3H]DPCPX and [ 3H1ZM241385 Binding Studies in Equine Forebrain

Saturation binding experiments for Al 1-selective ligand [3H]DPCPX and A2a-

selective ligand [3H]ZM241385 were performed in equine cortex and striatum

membranes to characterize Al and A2a binding sites in equine forebrain tissues. A best-

fit regression line suggested that [3H]DPCPX bound to a finite population of binding sites

in cortex with Kd and Bmax values of 0.29 nM and 972 + 24 fmole/mg protein,

respectively. In striatum, Kd and Bmax values were 0.58 nM and 580 + 15 fmnole/mg

protein, respectively (Fig. 3-1). On the other hand, [3H]1ZM241385 bound to a single class

of recognition sites in striatum with a Kd value of 0.9 nM and Bmax of 274 19

fmole/mg protein (Fig. 3-2). There was very minimal specific binding of [3H]ZM241385

to equine cortex membranes (data not shown). To further characterize that the binding of

[31H]DPCPX in striatum was A 1-selective, inhibition binding with the non-radioactive

A2a-selective compound ZM241385 was performed against [3H]ZM241385 binding in

striatum as well as against [3H]DPCPX binding in both cortex and striatum. The results

indicated that while ZM-241385 inhibition curves against [3H]DPCPX in cortex and

striatum were superimposable, the inhibition curve against [3H]ZM241385 in striatum

was markedly shifted to the left with Ki values 570-fold lower than that of [3H]DPCPX

(0.48 nM vs. 275 nM). In addition, it is clear from the graph that ZM241385 (50 nM)

blocked more than 95% of [3HJZM241385 binding in striatum while more than 90% of

[3H]-DPCPX binding in the same tissue was uninhibited (Fig. 3-3), indicating this is a

critical concentration point for ZM241385 to differentiate Al and A2a binding sites in

equine striatum.

Competition Binding of MX at Adenosine Receptors

In order to determine the antagonistic potencies of MX to Al and A2a receptors,

competition binding of MX against [3H]DPCPX to Al binding sites in cortex and against

[3H]ZM241385 binding to A2a binding sites in striatum were performed and antagonism

at respective adenosine receptors were compared.

Competition binding curves for CA, TP, TB and XN in Al and A2a adenosine

receptor binding sites are depicted in Fig. 3-4. The IC50 and Ki values are compared in

Table 3-1.

Except for TB, which has very low binding affinity at both receptor subtypes, Ki

values for CA, TP and TB all fell within low .M range. With regard to possible receptor

selectivity, MX were considered equal-potent at both receptor types despite CA and XN

were 2-fold more selective on A2a and TP was 2-fold more selective on Al receptors.

The rank order potency of binding affinities for MX was TP>XN>CA>>TB at both

classes of adenosine receptors.

Agonist and Antagonist Action on [35S]GTPyS Binding

In order to validate the use of [35S]GTPyS binding for the differentiation of

agonist from antagonist binding in equine adenosine receptors, adenosine receptor

agonist (adenosine) and non-xanthine antagonist (CGS 15943) were examined for their

stimulant effects on G-protein mediated [35S]GTPyS binding. The results indicated that,

in both cortex and striatum, adenosine increased [35SJGTPyS binding in a concentration-

dependent manner with maximal stimulation occurring at 10 [M adenosine and a

maximal increase of 53% in cortex and 158% in striatum over unstimulated controls

(Fig. 3-5). The non-xanthine adenosine antagonist CGS-15943 did not stimulate

[35S]GTPyS binding over the concentration range from 10 nM to 1 mM and it constantly

antagonized adenosine-induced [35S]GTPyS binding in a concentration-dependent manner

(Fig. 3-6). Therefore, we have demonstrated that binding of [35S]GTPyS in both tissues is

increased upon agonist-binding and is unaffected by antagonist-binding. Further

competition binding experiments were carried out in the same fashion to investigate the

effects of MX-binding on [35S]GTPyS stimulation. The results are depicted in Fig. 3-7

(for binding in cortex) and Fig. 3-8 (for binding in striatum). [35S]GTPyS binding was

either unchanged or slightly decreased (0 to 20%) by increasing concentrations of MX

ranging from 100 nM to 1 mM. In the presence of 4 uM adenosine, MX readily

antagonized adenosine-induced [35S]GTPyS binding. These results are consistent with

antagonist-binding properties and, therefore, suggest that CA, TP, TB and XN act as

antagonists at adenosine binding sites. Visual examination of the rank order potency of

MX in antagonizing adenosine-induced [35S]GTPyS stimulation was TP>XN>CA>>TB

in both cortex and striatum. The antagonist potencies determined here were in agreement

with their affinities to adenosine Al and A2a receptors insofar as MX with higher affinity

showed a stronger antagonistic potency.

The aim of this study was to pharmacologically characterize adenosine Al and

A2a binding sites in equine forebrain tissues and to examine the antagonistic properties

and potencies of selective MX that have medical importance and behavioral/performance

implications in the horses. To achieve this goal, radioligand binding studies were

performed in equine cortex and striatum membranes. The availability of high affinity,

selective radioligands has allowed pharmacological characterization of receptors with

more accuracy and has greatly facilitated the discovery of new receptor subtypes. In this

study, [311]DPCPX is by far the most widely used A 1-selective, xanthine based,

antagonist ligand with 500 to 700-fold selectivity for A2a receptors in rat whole brain

tissue (Bruns et al., 1987; Maemoto, et al., 1997). [3H]ZM241385, a commercially

available radioligand, is the most potent (Poucher et al., 1995; Keddie et al., 1996;

Muller et al., 2000) and A2a-selective (Ongini et al., 1999) antagonist ligand for the

study of adenosine receptors. Antagonist radioligands were used in view of their

advantages over agonist ligands that antagonists do not preferably bind to high affinity

states of the receptor, thus allowing more direct interpretation of competition curves.

With the selective antagonist ligands, equine cortex showed high affinity, high density

binding sites for A 1-selective ligand [3H]DPCPX (Kd/Bmax value of 0.29 nM / 972

fmole/mg protein) with very few A2a-selective [3H]ZM241385 binding sites identified.

On the other hand, equine striatum showed high affinity binding sites for both

[3H]DPCPX and [3H]ZM241385 with Kd/Bmax values of 0.58 nM / 580 fmole/mg

protein and 0.9 nM / 274 fmole/mg protein, respectively. The Kd values for [3H]DPCPX

binding in equine cortical tissue were in close agreement with previous findings in bovine

cortical membranes where a Kd and Bmax values of 0.22 nM and 1350 fmol/mg,

respectively, have been reported (Klotz, 1991). The Kd value for [3H]DPCPX binding

sites in striatum was also comparable to previous reports using the same ligand to label

whole brain tissues of other species including hamsters and rats (0.5 nM) as well as rabbit

and pig (0.75 nM), but was notably different from guinea pig (2.08) and sheep (0.17).

The affinity for A2a binding sites in different species is not as well established as

for Al sites. In contrast to a Kd/Bmax value of 0.48 nM / 1680 fmole/mg protein recently

reported for [3H]ZM241385 in rat striatum (Alexander and Millns, 2001), A2a binding

sites in equine striatum exhibited a slightly higher Kd and a significantly lower binding

capacity than the results from rats.

The fact that equine striatal tissue also contained Al binding sites is not

unprecedented since similar observations have been reported for rodents (Maemoto et al.,

1997; Laitinen, 1999). However, the high affinity and density ofAl binding sties in

equine striatum was surprising and therefore, further investigation on the selectivity of

[3H]DPCPX in equine striatum was conducted. Competition binding of the non-xanthine

antagonist ZM241385 against [3H]DPCPX and [3H]ZM241385 was performed in both

cortex and striatum. While the competition curves for [3JH]DPCPX in cortex and striatum

were virtually superimposable, the [3H]ZM241385 competition curve in striatum was

markedly shifted to the left by at least two orders of magnitude. The Ki ratio of

[3H]DPCPX /[3H]ZM241385 was 570, confirming the receptor selectivity of the ligand.

Further observation of the [3H]DPCPX competition curve revealed that binding inhibition

was not noted until ZM241385 concentration reached 50 nM, at which point more than

95% of [3H]ZM241385 binding was blocked, suggesting that ZM241385 could be useful

in improving separation of A 1-specific binding sites in equine striatum. In contrast to Al

binding sites in striatum, equine cortex exhibits very little A2a binding sites as defined by

the ligand [3H]ZM241385. Similar observations have been reported for several other

species regardless of ligand choice, although cortex A2a binding sites have been reported

in rat (Wan et al., 1990, Weaver, 1993, Cunha et al., 1994). In addition to species

differences, pharmacological characterization includes the use of various compounds to

identify a binding profile of each receptor subtype. Therefore, rank order potencies of

MX were also investigated by competition binding experiments in this study. The current

results indicated that despite the affinity of xanthine-based adenosine receptor antagonist

for [3H]DPCPX binding have been reported to be species-dependent (Maemoto et al.,

1997), The affinities (Ki) of CA, TP, TB and XN for both [3H]DPCPX (Al) and

[3H]ZM241385 (A2a) binding sites followed a common rank order potency of

TP>XN>CA>>TB as seen in rat, bovine brain and human platelet membranes (Schwabe,

1985) and various radioligands (Lohse et al., 1987; Jarvis et al., 1989). Selectivity

analysis revealed that while CA and XN were about two times more potent at A2a sites

([3H]ZM241385 in striatum) versus Al sites ([3 H]DPCPX in cortex), TP was twice as

potent at Al sites. However, such selectivities were not considered to be significant and

therefore we conclude that MX are equipotent at both adenosine receptor types in the

equine brain (Table 3-1). This is in agreement with other selectivity studies reported for

rat and bovine brain tissue with the agonist radioligand [3H]-N6-R-(phenylisopropyl)

adenosine ([3H]-PIA). In that study, CA was found to be more potent at Al sites whereas

TP was more potent at A2a sites (Schwabe et al., 1985).

The use of well established selective radioligands (Bruns et al., 1987; Alexander,

1999) and sensitive binding assay protocols for adenosine receptor experiments in

various animal species have ensured the consistency and reliability of this study and

greatly reduced the effort required to optimize assay conditions.

The use of ADA, DMSO and MgC12 in adenosine receptor binding assays was

among the most highly considered factors in the protocol. With few exceptions (Lohse et

al., 1987), the inclusion of 0.1-2 unit/mL ADA seems to be advantageous and has been

employed routinely for both Al and A2a binding assays (Bruns et al., 1986, 1987; Klotz

et al., 1991; Maemoto et al., 1997; Muller et al., 2000). DMSO was the other commonly

used compound contained in assays where higher drug solubility was required. Although

20% DMSO has been reported to decrease specific binding by 20% (Bruns et al., 1987),

5% was reportedly well tolerated (Muller et al., 2000) and 1% of DMSO (final

concentration) was most often used in the assay. Also commonly included especially in

A2a binding experiments was MgCl2 (10 mM), which has been shown to increase

radioligand binding and to reduce non-specific binding (Bruns et al., 1986). However, the

benefit was inconclusive (Zocchi, 1996; Nonaka, 1994; Muller, 2000). In the present

study, a final concentration of 0.1 to 1 unit/mL ADA and 1% DMSO were included but

MgCl2 was not employed. The presence of ADA and 1 %DMSO did not appear to show

any inhibitory effect on either [3H]DPCPX or [3H]ZM241385 binding while 10 mM

MgCl2 did not improve specific binding of [3H]ZM241385 in equine striatum (data not


Perhaps the most significant difference in the current protocol was incubation

time. A 1 to 2-hour incubation at 25C was regularly used for Al binding assays (Bruns

et al., 1987; Lohse et aL, 1987; Klotz et al., 1991; Maemoto et al., 1997), and for some

A2a binding assays (Maemoto, 1997; Jarvis et al., 1989). The current study required only

15 and 30 minutes incubation at room temperature to reach maximal binding for

[3H]DPCPX and [3H]ZM241385, respectively. In fact, maximal [3H]ZM241385 specific

binding started to decline when the incubation period exceeded 45 minutes (data not

shown). The reduction in the incubation time for the A2a receptor binding study may be

due in large part to the newer selective antagonist ligand. A common range of 60 to 90

minutes incubation period for agonist ligands 2-[p-(2-carboxyethyl)-phenethylamino]-5'-

N-ethylcarboxamino adenosine ([3H]-CGS-21680), [3H]-5'-N-ethyl-

carboxamidoadenosine ([3H]-NECA) or [3H]-PIA (Schwabe et al., 1980; 1985) have been

greatly decreased to 20 minutes by the antagonist ligand 3[H]-3-(3-hydroxypropyl)7-

methyl-8-(m-metnoxy-styoyl)-l-propargylxanthine ([3H]-MSX-2) and 30 minutes for

[3H]ZM241385 in this and other studies (Alexander and Millns, 2001). While the effect

of incubation time warrants more inspection on ligand association/disassociation across

species, shorter incubation time (20 minutes) has been recently adopted in [3H]DPCPX

binding in rat striatum (Maemoto, et al., 1997).

In determining the antagonistic properties of a compound, most often a non-

hydrolysable GTP-analog was included in a competition study and observed for the

presence of agonist-facilitated "shift" of receptor affinity state. This approach, however

while successful in distinguishing between Al adenosine agonists and antagonists using

the radioligand [3H]DPCPX (Van der Wenden et al., 1995), fails to distinguish antagonist

from agonists when antagonist ligands are used (Muller et al., 2000). Alternatively in this

study, the ability of MX to stimulate [35SJGTPyS binding to adenosine-receptor coupled

G-protein was used as a qualitative measurement of their antagonistic properties. The use

of [ S]GTPyS binding assay offers the advantage of studying receptor-G-protein

interactions independent of the effector system. While the effects of adenosine receptor

agonist in equine brain membranes is not the focus of the current study, it is important to

point out that our results for the agonist (adenosine) used in this study have revealed a

more than 3-fold maximal stimulation of [35S]GTPyS binding in the A2a-rich striatum

than in the A 1-rich cortex (presented as percentage over unstimulated controls). A2a-

receptor-mediated stimulation of [3 S]GTPyS binding has not been widely reported. We

found one previous study which suggested no evidence of A2a-receptor mediated

[35S]GTPyS stimulation in bovine cortex tissue (Lorenzen, 1993). However, the lack of

A2a-binding sites in bovine cortex may account for the lack of A2a-mediated [35S]GTPyS

stimulation in that study. In the present study, nearly equivalent amounts of cortical (70

.g/mg protein) and striatal (50 ug/mg protein) membranes were used in the assay.

Therefore the higher maximal stimulation in striatum could indicate a greater number or

more efficient G-protein coupling with adenosine receptor populations. Whether or not

the larger populations that contribute to a higher [35S]GTPyS binding included A2a

receptors remained unclear since high density Al binding sites were also found in

striatum. However, Al -mediated [35S]GTPyS binding in striatum alone may not fully

explain the higher maximal stimulation, given that the distribution of Al receptors and

binding capacity of [35S]GTPyS was expected to be equal, if not higher in cortex than in

striatum (Laitien, 1999; Moore et al., 2000). Kd and Bmax values for [35S]GTPyS in both

equine cortex and striatum Al receptors warrant further study. The use of A2a-selective

agonist or antagonist to stimulate or block [35S]GTPyS binding in striatum may provide

the most valuable piece to this puzzle. Nevertheless, based on current results, in view of

the almost 3 folds difference in maximal stimulation, it is reasonable to suggest a possible

role ofA2a-mediated [35S]GTPyS stimulation in equine striatum.

Another factor that should not be overlooked in the above discussion was the

presence of basal adenosine levels in the two tissues. The maximal [35S]GTPyS

stimulation was expressed as the percentage increase over basal levels (unstimulated

control). Therefore, a higher basal cortex adenosine level would be reflected as a higher

level of basal [35S]GTPyS binding which would tend to reduce the percent increase in

adenosine-stimulated [35S]GTPyS binding in cortex. The presence of endogenous

adenosine and hence basal adenosine-stimulated G-protein activation is supported by the

results that MX (as well as non-xanthine antagonists CGS15943) alone, decreased

[35S]GTPyS binding for up to 20% rather than having no effect on [35S]GTPyS binding as

would be expected from a typical antagonist binding. The influence of brain basal

adenosine levels on adenosine receptor-dependent [35S]GTPyS stimulation was not

thoroughly investigated until recently and was limited to Al-mediated pathway in rat

brain tissues (Laitinen, 1999; Moore et al., 2000). The extent to which A2a receptor was

involved in this process remains unclear. In the current study, efforts have been made to

reduce basal adenosine levels by an extra washing-step following preincubation of

membranes at 37 C for 20 minutes. Pre-incubating membranes at 37 C was expected to

facilitate the degradation of soluble adenosine as well as enhance the disassociation of

receptor-bound adenosine. However, the presence of reduced [35SJGTPyS binding in our

results indicates that endogenous adenosine was not completely eliminated by this

procedure. Inclusion of ADA in the assay would likely degrade endogenous adenosine

and reduce the signal/noise ratio. However, due to the fact that adenosine was used as

stimulating agent in this study, ADA was not employed in the assay. It should be noted

that the current study was initiated at a time when adenosine-dependent basal [35S]GTPyS

stimulation has not been widely investigated by techniques such as autoradiography and

our main focus was to "qualitatively" identify the antagonistic property of MX.

Therefore, the current design still suited our purpose of study. Even until today, only

limited study for Al-mediated [35S]GTPyS activation has been addressed, little is known

regarding A2a-mediated effects on [35S]GTPyS stimulation due to lack of suitable

selective A2a agonist (Ribeiro, 1999). In the future, choice ofagonist in similar study

should be cautioned. While basal adenosine levels have been proven to be a concern in

rat brain tissues, use of non-adenosine analog agonists should greatly reduce the conflict

situation. In addition, in the case of equine brain tissue, where Al and A2a receptors

coexist in high density in striatum, use of Al and A2a selective antagonists, DPCPX and

ZM241385 in combination with selective adenosine receptor agonists should facilitate the

understanding of the role of each receptor subtype in the activation of adenosine receptor-

bound G-protein as determined by [35S]GTPyS bindings.

In conclusion, we have identified high affinity, high-density adenosine A I

binding sites in equine cortex and striatum as well as A2a binding sites in equine

striatum. Pharmacological characterization of these binding sites reveal that CA and its

metabolites exhibit a rank order potency that was identical at both receptor types and

similar to their affinities at adenosine receptors in other species. The receptor selectivities

of CA, TP and XN were slightly different in equine brain relative to bovine or rat brain,

but was considered equal potent at both receptor types. The determination of affinities of

CA and its metabolites in equine adenosine Al and A2a receptors should facilitate the

interpretation of pharmacokinetic studies of these compounds in CNS and their roles in

behavioral stimulations of horses.

Table 3-1. Competition of MX for [3H]DPCPX binding in equine cortex and
[3H]ZM241385 binding in equine striatum membranes. IC50 values shown are mean
( SEM) from 3 to 4 experiments. Ki values are derived from IC50 using Cheng-Prusoff
equation and present as geometric mean with range from 3 to 4 experiments.



Antagonist IC50o* Ki IC50so Ki

Caffeine 344 44 77(64-87) 178 28 38(29-56)
Theophylline 31 2 7(6-8) 76 18 16(12-27)
Theobromine 934 98 209 (190-240) > 10000 ND
Paraxanthine 153 5 35(33-38) 104 13 22(19-25)

*Values for IC5o and Ki are listed in micromolar (gM)
ND: Not determined

Total binding 1000
n 800
Specific binding 600

Non-specific bindin 400
r 200
D -

2 4 6 8 10 0 2

0 200 400 600 800 1000






Total binding

Specific binding
Non-specific binding-

4 6 8 10

0 150 300 450 600


Fig. 3-1. Representative saturation binding curve and Scatchard plot for [3H]DPCPX in
membranes from equine cortex (left panels) and striatum (right panels). Saturation
binding was carried out in assay buffer (50 mM Tris-HCl, pH 7.4, 0.1 unit/mL adenosine
deaminase, 1% DMSO) containing either 30 to 40 lag (cortex) or 40 to 50 u.g (striatum)
protein and 8 to 10 concentrations of [3HjDPCPX (0.05 to 10 nM) in a final volume of
500 gL as described in materials and methods. Best-fit pharmacological parameter
determined from linear regression revealed a Kd value of 0.29 (0.26 to 0.34) nM and
maximum binding capacity (Bmax) of 972 24 fmole/mg protein for cortex and a Kd of
0.58 (0.54 to 0.62) nM and a Bmax of 580 15 fmiole/mg protein for striatum. Points
shown are mean ( SEM) of duplicates from 4 experiments.

" 600
,- 300
* 0

Total binding -
600 0
400 Non-specific^
400 ,^bind'n^^

200 /^ ~^ '' tc if
20 Specific binding

0 4 8 12 16 20

[3H-ZM241385] (nM)


'" j

0 50 100 150 200 250 300


Fig. 3-2. Representative saturation binding curve and Scatchard plot for [3H]ZM241385
in equine striatum membranes. Saturation binding was carried out in assay buffer (50
mM-Ttris-HCI containing 120 mM NaCI and 5 mM KCI, pH 7.4, 1 unit/mL ADA, 1%
DMSO) containing 300 to 500 kig membrane protein and 10 concentrations (0.05 to 20
nM) of [3H]ZM241385 in a final volume of 500 gL as described in materials and
methods. Best-fit pharmacological parameter determined from linear regression revealed
a Kd value of 0.9 (0.69 to 1.09) nM and Bmax of 274 19 fmole/mg protein. Points
shown are mean ( SEM) of duplicates from 3 experiments.

in cortex
in striatum

in striatum

0.1 1 10 100 1000

[ZM-2413851 (nM)

Fig. 3-3. Inhibition binding curve for ZM241385 against [3H]DPCPX and
[3H]ZM241385 in equine cortex and/or striatum membranes. Incubation conditions are
identical to respective ligand binding assays as described in material and methods, except
for that 10 to 16 different concentrations of ZM241385 ranging from 0.1 nM to 10 pM
were included in assay. The Ki values for [3H]DPCPX in cortex and striatum were 214
and 275 nM, and for [3H]ZM241385 in striatum was 0.48 nM. Non-specific binding was
determined with 5 pM of CGS 15943. Points shown are mean ( SEM) of duplicates from
3 experiments.


(A) Cortex

.0 100


U 60
S* Caffeine
S *40 Theophylline
e- 20 Theobromine \,
20 Paraxanthine
0 -_
1 10 100 1000
(B) Striatum



0 60



1 0--
1 10 100 1000

[Methylxanthine] (p.M)

Fig. 3-4. Inhibition binding curves for MX against (A) [3H]DPCPX in equine cortex and
(B) [3H]ZM241385 in equine striatum membranes. A fixed concentration of [3HjDPCPX
(1 nM ).or [3H]ZM241385 (3 nM) was incubated with 9 log-spaced concentrations (0.1
jiM to 10 mM) of competing MX (CA, TP, TB, XN) in respective assay buffer as
described in materials and methods. Non-specific binding was determined with 5 pLM of
CGS 15943. Curves were generated using the best-fit parameter values from the logistic
equation. Points shown are mean ( SEM) of duplicates from at least 3 experiments.

(A) Cortex


0.001 0.01 0.1

1 10

0.001 0.01 0.1

[Adenosine] (LM)

Fig. 3-5. Adenosine-stimulated [35S]GTPyS binding in equine (A) cortex and (B)
striatum membranes. Membranes (70 ug protein cortex or 50 jig protein striatum) were
incubated with increasing concentrations of adenosine (0.2 nM to 15 gM) in a assay
buffer containing 50 mM Tris-HCl pH 7.4, 80 mM NaCI, 4 mM MgCI2, 1 mM EDTA, 1
mM dithiothreitol, 10 jM GDP, 0.5% BSA and 0.2 nM [35S]GTPyS. Non-specific
binding was determined with 20 gM of GTPyS. Basal level is defined as specific
[35S]GTPyS binding without the presence of adenosine. Points shown are mean ( SEM)
of duplicates on at least 3 experiments.


.5 E 120

| 80


1 0

(B) Striatum






1 10


0.01 0.1 1 10 100 1000

20 1__
0.01 0.1

1 10 100 1000

[CGS15943] (gM)

Fig. 3-6. CGS15943 antagonism of adenosine-stimulated 35SGTP yS binding to equine
(A) cortex and (B) striatum membranes. [35S]GTP yS (0.2 nM) binding was determined
by incubating CGS15943 (empty circle, 3 concentrations) or CGS 15943 in the presence
of 4 tM adenosine (filled circle, 100 nM to 1 mM at 5 concentrations) with assay buffer
and membrane preparations described in Fig. 3-5. Non-specific binding was determined
with 20 uM of GTPyS. Maximal level of stimulation is defined as specific [35SJGTPyS
bindings in the presence of 4 jiM adenosine and without antagonist. Points shown are
mean ( SEM) of duplicates on 3 experiments.