Introduction

Top Abstract Introduction Procedures Results Discussion References

One of the most destructive aspects of AIDS is an encephalopathy (AIDS dementia complex) associated with infection of the brain by the causative pathogen HIV-1 (Brenneman et al., 1988; Ho et al., 1985; Pang et al., 1990; Merrill and Chen, 1991; Wiley et al., 1986). The passage across the BBB and blood-CSF barrier to gain entrance into the CNS is therefore an important consideration in the evaluation of potential anti-AIDS drugs. The first drug to show some promise in treating AIDS by its ability to inhibit HIV replication in cell culture was the deoxyribonucleoside AZT (zidovudine, Retrovir; Mitsuya et al., 1985). It was transferred from the laboratory to a clinical setting with remarkable speed, and several controlled trials have shown that it reduces morbidity and mortality associated with the disease (Kahn et al., 1992; Yarchoan et al., 1986, 1987). Studies on the mode of action of AZT have shown that it is phosphorylated intracellularly to its 5-triphosphate form; it is this active form of AZT that can selectively inhibit HIV reverse transcriptase and incorporate itself into viral DNA, resulting in the termination of DNA chain elongation (Furman et al., 1986).

Several studies have concentrated on the pharmacokinetic distribution of radiolabeled AZT after systemic administration and found that although the drug was distributed in all tissues, the levels in the brain were comparatively low (Ahmed et al., 1991; De Miranda et al., 1990). In vivo studies of the brain uptake of AZT after a single pass through the cerebral circulation have also failed to demonstrate a significant transport of AZT across the BBB (Ellison et al., 1988; Terasaki and Pardridge, 1988). However, studies measuring concentrations in human CSF after AZT administration have produced CSF-to-plasma ratios of ~60% (range, 10-156%) (Cload, 1989; Klecker et al., 1987; Yarchoan et al., 1986, 1987, 1989). This, together with the observed improvement in neurological function of AIDS patients undergoing AZT treatment, has led to the conclusion that if AZT cannot cross the BBB, it must cross the blood-CSF barrier at the level of the choroid plexuses to enter the CNS (Ellison et al., 1988; Galinsky et al., 1990; Terasaki and Pardridge, 1988). Thus, AZT was believed to be similar to its parent compound, thymidine, in its route of entry into the CNS.

However, our research group recently identified a saturable uptake system for thymidine at the BBB by means of an in situ brain perfusion technique, which allows examination of the uptake of slowly permeating molecules into the CNS (Thomas and Segal, 1996). Furthermore, previous studies have demonstrated that the active nucleoside transport system in rabbit choroid plexus will not transport synthetic ribonucleosides or deoxyribonucleosides modified in the 2, 3 or 5 position (Spector, 1982b; Spector and Huntoon, 1984). This has been further confirmed using rabbit choroid plexus slices in which analogs modified on the ribose ring, such as AZT, dideoxycytidine, dideoxyadenosine and cytidine arabinoside, did not inhibit Na⁺-dependent nucleoside uptake (Wu et al., 1992, 1994). Thus, unlike thymidine, AZT cannot utilize the sodium-dependent nucleoside transporter thought to be present at the choroid plexus. Another important consideration is that if the choroid plexus is the only route for AZT entry into the CNS, then the levels of drug in the CSF will not necessarily reflect levels in the brain tissue. This is because the rate of bulk flow and drainage of the CSF through the arachnoid villi is thought to be faster than the diffusion kinetics of the drug from the CSF into the brain (Blasberg et al., 1975; Collins and Dedrick, 1983). If this is correct, then why is their substantial clinical improvement thought to be accounted for by AZT action on HIV-infected sites within the brain (Yarchoan et al., 1989; Maehlen et al., 1995)? Terasaki and Pardridge (1988) suggested that AZT may have a predominantly indirect effect on HIV levels within the CNS, with the drug either blocking meningeal proliferation of the virus or decreasing the concentration of the virus within the blood-borne mononuclear cells that enter the brain. It must also be mentioned that AIDS patients often have neurological problems associated with cerebrovascular complications (Smith et al., 1990; Snider et al., 1983). For example, increases in vascular permeability are thought to cause the subcortical cerebral atrophy associated with AIDS dementia (Power et al., 1993). The neurological improvement associated with AZT and the AZT levels observed in the CSF may be results of this increase in BBB permeability in AIDS patients. However, it is also possible that AZT may actually cross the BBB in some species, including humans (Yarchoan et al., 1989) and dogs (Gallo et al., 1992), but not in others, such as the rat (Terasaki and Pardridge, 1988), although a preliminary BUI study has suggested that AZT can cross the BBB of 6- to 8-week-old rats (Ayre et al., 1989).

These conflicting reports of AZT transport into and within the brain appear to only hamper further developments in the treatment of AIDS. Furthermore, it is not known whether AZT can use the saturable uptake system at the BBB for thymidine. The aims of the present study were to clarify the passage of AZT across the BBB and blood-CSF barrier and relate it to that of its parent compound, thymidine, thus achieving a clearer understanding of nucleoside analog movement into and within the CNS and aiding the clinical assessment of the mode of action, efficacy and optimal administration of AZT.

	Experimental Procedures

Top Abstract Introduction Procedures Results Discussion References

All experimental procedures were within the guidelines of the Animals (Scientific Procedures) Act 1986, UK.

Bilateral In Situ Brain Perfusion

The bilateral in situ brain perfusion method allows quantification of the transport of slowly moving molecules from the blood into both brain and CSF. The technique has been fully described previously (Thomas and Segal, 1996) and is only briefly detailed here.

Adult Dunkin-Hartley guinea pigs (250-500 g) were sedated with fentanyl/fluanisone (1 ml/kg i.m.), anesthetized with midazolam hydrochloride (1 ml/kg i.m.) and then heparinized (10,000 units/kg i.p.). The neck vessels were exposed, and the right common carotid artery was cannulated with fine polythene tubing connected to a perfusion system. At the start of perfusion, the right jugular vein was severed to allow drainage of the perfusion medium. Once the correct perfusion pressure of ~100 mm Hg and a perfusion flow rate of 3.7 ml/min had been achieved for the right carotid artery, the left carotid artery was cannulated and perfused in a similar manner. The remaining jugular vein was then severed. The perfusion medium consisted of oxygenated (95% O₂/5% CO₂) sheep erythrocytes suspended in a saline/dextran medium to a hematocrit of 20%. The [³H]AZT (8.2 nM) in the absence or presence of unlabeled AZT (5 mM) or D-[¹⁴C]mannitol (1.9 µM; plasma space marker) was then infused into the inflowing perfusion medium.

After the set perfusion time (2.5-30 min), a cisterna magna CSF sample (~50 µl) was taken, and perfusion was terminated by decapitation of the animal. The brain was removed, the choroid plexuses were excised and the cerebellum was separated from the cerebrum and homogenized separately. Perfusion fluid was collected from the carotid cannulae immediately after the end of the perfusion period. The perfusion fluid was centrifuged, and triplicate "plasma" perfusate samples were taken. Brain (~50 mg wet weight), CSF and plasma samples were then prepared for scintillation counting as described below.

Expression of Results

The amount of [³H] or [¹⁴C] radioactivity in the brain and CSF (C_Tissue; dpm/g or dpm/ml) was expressed as a percentage of that in the plasma perfusate (C_pl; dpm/ml) and termed R_Tissue (%). The unidirectional transfer constant, K_in, was then determined from the multiple-time uptake data (2.5-30 min) by use of the following equation: R_Tissue (T) = K_in · T + V_i, where T is the time of perfusion (min) and V_i is the initial volume of distribution of the test solute in the rapidly equilibrating space (Thomas and Segal, 1996). This equation defines a straight line with a slope K_in (ml/min/g) and ordinate intercept V_i (ml/g), which was determined by least-squares linear regression analysis of the multiple-time uptake data. Transport of the test solute from brain to blood is indicated by a departure from linearity of the experimental points. The K_in value for transport from blood to CSF was determined by single-time point uptake analysis (K_in = R_Tissue/T; Williams et al., 1996). Blood-to-brain K_in values could also be determined by single-time point analysis by initially correcting for vascular space by subtracting the R_Tissue value for D-[¹⁴C]mannitol from that for [³H]AZT.

Ventriculocisternal Perfusion

The ventriculocisternal brain perfusion technique allows investigations into the exchanges of substances among the CSF, brain and blood (Davson et al., 1982; Thomas et al., 1997).

Adult New Zealand White rabbits of either sex and 2.7- to 3.2-kg weight were anesthetized with bolus injections of fentanyl/fluanisone (0.3 ml/kg i.m.) and midazolam (0.2 ml/kg i.v.), and anesthesia was maintained with subsequent doses of fentanyl/fluanisone (0.15 ml/kg i.m.) and midazolam (0.2 ml/kg i.v.). Both lateral ventricles were then cannulated, and ventriculocisternal perfusion carried out as previously described by Davson et al. (1982). Each lateral ventricle was perfused at a rate of 33 µl/min with an artificial CSF (Thomas et al., 1997) containing [³H]AZT (0.17 µM) and blue dextran (0.005 g/ml), and samples were collected over 10-min periods via a cannula placed in the cisterna magna. The cisterna magna was kept at a negative pressure by applying suction to the outflow cannula via a vacuum pump (5 cm H₂O). This ensured that the flow of artificial CSF was kept within the ventricular system. After a 2-hr perfusion period, the animal was decapitated, and the brain and choroid plexuses were removed and immediately weighed. Due to its large molecular weight (2 × 10⁶), blue dextran is not appreciably lost from the ventricular system, which enables determination of the CSF secretion rate because its concentration is diluted with newly secreted CSF [i.e., CSF secretion rate equals the difference between the concentration of blue dextran in the entering and emerging perfusion fluids divided by that in the entering perfusion fluid multiplied by the perfusion fluid rate (66 µl/min)]. The concentration of blue dextran in perfusion fluid samples was determined using a SP30 series spectrophotometer (wavelength, 605 nm; Pye Unicam Ltd., Cambridge, UK). Duplicate samples were prepared for analysis by dilution with distilled water (1:10). Samples were also prepared for scintillation counting as described below.

Expression of Results

Clearance. The clearance or the volume of perfusion fluid, of average concentration, cleared of [³H]AZT per minute (Bradbury and Davson, 1964) was determined with the following equation: where C_in is the radioactivity per unit volume of entering perfusion fluid, C_out is the radioactivity per unit volume of emerging perfusion fluid, F_in is the rate of perfusion (µl/min) and R_D is the steady-state value of C_in/C_out for blue dextran.

R_BR, R_CP and R_CSF. The uptake of [³H]AZT into the brain (R_BR) or choroid plexus (R_CP) was expressed as a percentage ratio of that found in the perfusion fluid, and the R_CSF was determined as the mean value of (C_out/C_in) × 100 derived from the last four samples of the perfusion fluid.

Losses to blood and brain. The total loss of radioactivity is given by the difference between the amount perfused (F_in × C_in × time) and the amount recovered (amount in collected perfusate during 120 min, _{0-120 min} C_out × time). The amount recovered in brain is calculated from R_BR and brain weight, and the amount lost to the blood is obtained by the difference between the total lost and that found in the brain. To allow comparison from experiment to experiment, the losses were calculated by setting the activity in the perfusion fluid arbitrarily equal to 100.

Radioactive Counting

Brain, CSF, perfusion fluid (100 µl) and choroid plexus samples were treated similarly and solubilized overnight in 0.5 ml of Soluene-350 (Canberra-Packard Ltd., Pangbourne, UK). Before isotopic counting, 3 ml of scintillation fluid was added, and the [³H] and/or [¹⁴C] activities estimated using an LKB Spectral beta liquid scintillation spectrometer and, with the use of internal stored quench curves, converted to disintegrations per minute. All results were corrected for background.

Octanol/Saline Partition Coefficient

The partition coefficient for [³H]AZT in 1-octanol/saline was determined according to the method previously described (Thomas and Segal, 1996).

Materials

Both [methyl-³H]AZT (specific activity, 9.1 Ci/mmol), which was custom-synthesized by Sigma Chemical Co. (St. Louis, MO), and AZT (molecular weight, 267.2) were kind gifts of the The Wellcome Foundation Ltd. (Beckenham, Kent, UK). D-[¹⁴C]Mannitol (specific activity, 32 mCi/mmol) was purchased from ICN Radiochemicals (High Wycombe, Bucks, UK). Radiochemical purity was reported as 95% by the companies for both isotopes. Unless specified, all other reagents and chemicals were obtained from Sigma Chemical Co. (Poole, UK).

Statistical Analysis

Values are reported as mean ± S.E.M. Student's t test (paired or unpaired as appropriate) was used for the comparison of two means and the level of significance taken as P < .05.

	Results

Top Abstract Introduction Procedures Results Discussion References

Figure 1 illustrates the progressive uptake of [³H]AZT into the cerebrum, cerebellum and CSF plotted as a function of time. The uptake of [³H]AZT into the cerebrum was initially 1.85 ± 0.36% at 2.5 min and rose to 4.03 ± 0.44% at 30 min and was significantly different than the comparative uptake of D-[¹⁴C]mannitol (P < .001). In addition, the uptake of [³H]AZT into the cerebellum was 2.08 ± 0.52% at 2.5 min and rose to 3.87 ± 0.66% at 30 min, which was significantly different than the uptake of D-[¹⁴C]mannitol (P < .01). The uptakes of [³H]AZT into the cerebrum and cerebellum were not significantly different from each other. The entry of [³H]AZT into CSF was 0.81 ± 0.29% at 2.5 min and rose to 2.87 ± 0.26% at 30 min (fig. 1). Although the uptake measured at 10 min was obviously not significantly different than that of D-[¹⁴C]mannitol, [³H]AZT uptakes measured at 2.5, 20 and 30 min were significantly different than that of D-[¹⁴C]mannitol at the same time periods (P < .01). The uptake of [³H]AZT into the cerebrum or the cerebellum, once D-[¹⁴C]mannitol/vascular space had been considered, was not significantly different than that into the CSF. The K_in and V_i values determined for [³H]AZT entry into the cerebrum, cerebellum and CSF were similar to each other, as were the K_in values determined by multiple- and single-time uptake analyses (table 1). However, although the V_i values for [³H]AZT were similar to those previously determined for [³H]thymidine, K_in values were ~3-fold higher for [³H]thymidine (Thomas and Segal, 1996). Figure 2 illustrates the effect of an excess of unlabeled AZT on the uptake of [³H]AZT into cerebrum, cerebellum and CSF. As can be seen, there was no significant reduction in [³H]AZT uptake into any of the tissues measured.

View larger version (15K): [in this window] [in a new window]   Fig. 1.   Blood-to-CNS uptake of [3H]AZT and D-[14C]mannitol plotted against time measured using the bilateral in situ brain perfusion technique. Uptake is expressed as the percentage ratio of tissue to plasma activities (RTissue ml/g). Each point represents the mean ± S.E.M. of 4-6 animals. Kin and Vi values were determined as the slope and ordinate intercept of the computed regression lines.

View larger version (18K): [in this window] [in a new window]   Fig. 2.   The effect of an excess of unlabeled AZT on the blood-to-CNS uptake of [3H]AZT measured using the bilateral in situ brain perfusion technique. Uptake is expressed as the percentage ratio of tissue to plasma activities (RTissue ml/g), and D-[14C]mannitol/vascular space has been subtracted from the brain values. Values are the mean ± S.E.M. for 4-6 animals. Unpaired Student's t test was used to compare the [3H]AZT uptake values in the absence and presence of unlabeled AZT; in all cases, P > .05.

The effect of perfusing an artificial CSF containing [³H]AZT and the large-molecular-weight molecule blue dextran through the ventricles of the rabbit is illustrated in figure 3. Although the ratio of recovered AZT was considerably less than that of blue dextran, both compounds can be seen to reach a steady state after 70 min. The rate of CSF secretion was calculated from the mean of the last four outflow samples for blue dextran and found to be 6.8 µl/min, which is similar to values previously reported (Thomas et al., 1997). Table 2 shows the R_CSF, clearance, R_BR and R_CP values determined for [³H]AZT and, for comparison purposes, the previously published values for [³H]thymidine (Thomas et al., 1997). Although there was no significant difference between the R_BR values for [³H]AZT and [³H]thymidine, the R_CSF, clearance and R_CP values for [³H]AZT were significantly different than those for [³H]thymidine. Table 2 also shows the percentage loss of [³H]AZT from the CSF perfusion fluid and the proportion of this total loss, which enters the brain and blood. In contrast to those values determined for [³H]thymidine, it can be seen that the majority (79.9%) of [³H]AZT remains in the ventricular system. However, the percentage of [³H]AZT that enters the brain is similar to that previously determined for [³H]thymidine. The octanol-saline partition coefficient determined for [³H]AZT was 1.0195 ± 0.0233. 

View larger version (17K): [in this window] [in a new window]   Fig. 3.   Ventriculocisternal perfusion in the rabbit with an artificial CSF containing blue dextran () and [3H]AZT (). The percentage ratio of the concentrations found in the emerging perfusion fluid vs. that found in the entering perfusion fluid is plotted as a function of time in min. Values are mean ± S.E.M. for 3 animals at the midpoint of 10-min collection periods.

	Discussion

Top Abstract Introduction Procedures Results Discussion References

The uptake of [³H]AZT into the CNS was measured using the bilateral perfusion technique over 30 min in the anesthetized guinea pig and found to be significantly greater than the uptake of the inert polar molecule D-[¹⁴C]mannitol (fig. 1). This suggests that [³H]AZT can measurably cross the BBB and blood-CSF barrier and enter the brain and CSF. In contrast to these results, the brain uptake of [³H]AZT measured in earlier studies using the single-pass technique (<30 sec) was extremely low and not significantly higher than the brain extraction of [¹⁴C]sucrose that does not measurably cross the BBB (Ellison et al., 1988; Terasaki and Pardridge, 1988). Several other studies have also suggested that AZT cannot significantly cross the BBB (Ahmed et al., 1991; De Miranda et al., 1990; Galinsky et al., 1990) and can only enter the brain via the choroid plexus/CSF pathway (Ellison et al., 1988; Galinsky et al., 1990; Terasaki and Pardridge, 1988). It is interesting to note that using the single-pass technique, thymidine (the parent compound of AZT) was also believed not to cross the BBB but rather to enter the brain via the choroid plexus/CSF pathway (Cornford and Oldendorf, 1975; Spector, 1982a); it was not until the development of techniques that could study the passage of more slowly moving molecules that a saturable transport system at the BBB for this pyrimidine deoxyribonucleoside was indicated (Thomas and Segal, 1996).

The unidirectional transfer constants (K_in) for [³H]AZT uptake into the cerebrum, cerebellum and CSF from the blood were similar to each other (table 1), so the homogeneous level of [³H]activity observed in the brain (fig. 1) is unlikely to be a result of transport predominantly through the choroid plexus/CSF pathway. This can be concluded on the basis of several facts concerning blood, brain and CSF interactions and the ventriculocisternal perfusion results illustrated in figure 3 and table 2. First, the surface area of the choroid plexuses are significantly smaller than that of the BBB (Keep and Jones, 1990). Furthermore, as shown in table 2, the CSF is more likely to act as a sink to the brain than is the brain to act as a sink to the CSF because the rate of bulk flow out of the ventricles is considered to be faster than the diffusion kinetics from this fluid into the brain (Collins and Dedrick, 1983; Davson et al., 1961) and there is a very slow turnover of brain extracellular fluid (Cserr et al., 1981; Szentesvanyi et al., 1984). Thus, if the choroid plexus route was the exclusive source of a substance, high- solute concentrations may be produced at the ependymal-CSF surface, but there would be a sharp decease in tissue concentrations as close as 0.2 cm to the surface (Blasberg et al., 1975). Although the relative contribution of the BBB or choroid plexus pathway to the level of [³H]AZT found in the CSF after in situ brain perfusion experiments cannot be determined, for the reasons outlined previously and the results shown in table 2, the BBB must play an important role in supplying the brain and CSF with [³H]AZT. A pharmacokinetic study in which brain and CSF concentrations were compared after intra-arterial and intravenous infusions of AZT also supports the ability of AZT to enter brain parenchyma at concentrations similar to those in the CSF (Gallo et al., 1992).

Although a range of AZT dosage regimens for the management of patients with HIV infection has been used, 500 or 600 mg/day in two to five divided oral doses has been commonly used worldwide (Cooper et al., 1993). A Phase 1 study has shown that the plasma concentration after the oral administration of AZT (5 mg/kg) ranges from 0.4 to 7.1 µM. The K_in values determined from the in situ brain perfusion experiments of the present study (table 1) suggest that if the plasma concentration were in this range, the influx of AZT into the CNS of the guinea pig would correspond to 0.4 to 6.6 pmol/min/g. The relationship between in vitro susceptibility of HIV to AZT and clinical response to therapy remains under investigation, but in vitro pharmacodynamic studies have indicated that continuous exposure to AZT concentrations of 1 µmol/l are required to optimally suppress HIV infection (Balis et al., 1989). If we assume that the concentration of AZT in the guinea pig plasma has reached 7.1 µM, then it would take ~2.5 hr, with no removal or metabolic processes occurring, to reach this critical concentration of 1 µmol/l in the guinea pig brain.

The octanol/saline partition coefficient for [³H]AZT determined in the present study was 1.02 ± 0.02 and is similar to the previously reported values of 1.26 (Zimmerman et al., 1987) and 1.10 ± 0.04 (Collins et al., 1988). It appears that replacement of the 3-hydroxyl group of thymidine with the 3-azido moiety to form AZT (fig. 4) confers substantial lipophilicity to AZT compared with thymidine itself (octanol/saline partition coefficient, 0.0572 ± 0.0017; Thomas and Segal, 1996). However, although the lipophilicity of [³H]AZT (molecular weight, 267.2) was found to be >17-fold greater than that of [³H]thymidine (molecular weight, 242.2) and the molecules are of a similar size, the results determined for [³H]AZT after both in situ brain and ventriculocisternal perfusions were considerably lower (K_in, R_CP, clearance) or similar (V_i, R_Br) to the values determined for [³H]thymidine (table 1; Thomas and Segal, 1996).

View larger version (10K): [in this window] [in a new window]   Fig. 4.   Comparative structures of AZT and thymidine. AZT is an analog of thymidine in which the 3-hydroxyl group is substituted with an azido group.

The general topic of entry of drugs into the CNS has been investigated extensively, and the principal determinants, especially lipophilicity and protein binding, are well characterized (Davson and Segal, 1996). However, there are deviations from these general principles, and as this study demonstrates, the lipid solubility of a solute is not an infallible guide to its potential ability to penetrate the BBB and blood-CSF barrier. To investigate whether the measured uptake of [³H]AZT into the cerebrum, cerebellum and CSF was via a carrier-mediated transport system, several experiments determined the effect of an excess of unlabeled AZT on the uptake of [³H]AZT (fig. 2). These results indicated that in contrast to its parent compound, thymidine (Thomas and Segal, 1996), AZT did not use a saturable mechanism to enter the CNS but instead entered both brain and CSF purely by a process of diffusion. This would therefore explain the lower values determined for [³H]AZT compared with [³H]thymidine (table 1). Studies of the transport into the CSF from the plasma, as well as passage across the in vitro BBB and the plasma membranes of human erythrocytes, lymphocytes and cultured intestinal epithelium, have also shown that AZT does not use a nucleoside transport system but instead crosses the membrane by nonfacilitated diffusion (Hu, 1993; Masereeuw et al., 1994; Zimmerman et al., 1987). The pharmacological distribution and metabolism of AZT in mice have also been shown not to follow that of the parent molecule, thymidine (Ahmed et al., 1991).

Among nucleoside analogs, AZT is highly unusual in its apparent ability to permeate cell membranes totally by nonfacilitated diffusion (Domin et al., 1992; Paterson et al., 1981; Zimmerman et al., 1987). This somewhat unique characteristic of AZT may be attributed largely to its substantial lipophilicity but also to the structural modifications tolerated by the nucleoside transporter. Because all natural purine and pyrimidine nucleosides, although not with equal efficiency, are thought to be transported by the facilitated nucleoside transporter, the size of the carbon 1 substituent is not of major importance (Plagemann et al., 1988). However, modification of the sugar moiety is thought to be more critical (Brett et al., 1993; Cass and Paterson, 1972, 1973; Gutierrez and Giacomini, 1993). Considerable flexibility must exist in recognition of the sugar moiety of nucleosides because both ribonucleosides and deoxyribonucleosides are accepted by the carrier (Gati et al., 1984; Taube and Berlin, 1972). However, it has been observed that modification at the 3-position of the sugar greatly reduces nucleoside interaction with the murine erythrocyte (Gati et al., 1984) and the rabbit leukocyte nucleoside transporters (Taube and Berlin, 1972). Because AZT is formed by replacement of the 3-hydroxyl group of thymidine with an azido moiety, this circumstance may serve to exclude AZT from cell permeation via this carrier (fig. 2). Previous studies have demonstrated that the active nucleoside transport system in rabbit choroid plexus will transport thymidine but not synthetic ribonucleosides or deoxyribonucleosides modified in the 2, 3 or 5 position (Spector, 1982b; Spector and Huntoon, 1984; Wu et al., 1992, 1994). This fact might be the reason for the significantly lower levels of accumulated [³H]AZT compared with [³H]thymidine observed within the choroid plexus tissue isolated after ventriculocisternal perfusion experiments in the rabbit (R_CP; table 2), although probenecid, a competitive inhibitor of the active transport system for weak acids from the CSF to blood, has been shown to inhibit AZT elimination from the CSF in rabbits (Hedaya and Sawchuk, 1989).

It is known that anesthetics have effects on cerebral blood flow and metabolism (Goldman and Sapirstein, 1973). Thus, because other investigators have shown a general linkage among BBB transport, blood flow and brain metabolism (Braun et al., 1985), it is likely that there is a similar depressive effect on BBB substrate transport. Therefore, nucleoside transport measured in anesthetized animals is not necessarily the same as nucleoside transport in conscious animals. The sedatives/anesthetics used in this study included a morphine analgesic, fentanyl, and a benzodiazepine anesthetic, midazolam. Their ability to inhibit the binding of the equilibrative nucleoside transport inhibitor NBMPR has previously been investigated; it has been shown that although midazolam inhibits NBMPR binding in guinea pig cortical membranes, fentanyl does not (Hammond and Clanachan, 1984). It is thus important to remember that the uptake of AZT and thymidine (Thomas and Segal, 1996) into the CNS is possibly reduced due to the effects of the benzodiazepine anesthetic midazolam. However, it has also been suggested that the benzodiazepines are very weak inhibitors of NBMPR binding in guinea pig and rat CNS membranes and therefore are unlikely to affect nucleoside uptake at therapeutic levels (Hammond and Clanachan, 1984; Marangos et al., 1982).

In summary, this present study indicates that AZT can cross the BBB and blood-CSF barrier and gain entry into the CNS but only by a diffusive process. The replacement of the 3-hydroxyl group of thymidine with the azido moiety to form AZT appears to increase the lipophilicity but at the same time reduces the ability of the nucleoside to be carried by the nucleoside carrier. This study thus confirms the importance of the 3-hydroxyl position on the sugar part of the nucleoside as a site of substrate recognition by the nucleoside carrier and indicates that AZT does not follow its physiological counterpart in its mode of passage across membranes.