Drug-Protein Binding and Blood-Brain Barrier Permeability

Experimental Procedures

Materials.

[¹⁴C]Diazepam was obtained from Amersham International (Buckinghamshire, UK). [³H]Diazepam, [¹⁴C]caffeine,l-[³H]propranolol, [³H]glucose, [¹⁴C]sucrose, [³H]sucrose, [³H]water, and [¹⁴C]butanol were obtained from NEN Life Science (Boston, MA). (+)-[¹⁴C]S-145 Na [(+)-[1R-[1a,2a(Z),3b,4a]]7-[3-[[U-¹⁴C]phenylsulfonyl)amino]bicyclo[2.2.1]hept-2-yl]5-heptenoic acid sodium salt], [¹⁴C]S-312-d [(S)-(+)-methyl 3-isobutyl6-methyl-4-(3-nitrophenyl)-4,7-dihydrothieno[2,3-b][4-¹⁴C]pyridine5-carboxylate], and [¹⁴C]S-8510 [2-(isoxazol-3-yl)-3,6,7,9-tetrahydroimidazo[4,5-d]pyrano[4,3-b][4-¹⁴C]pyridine monophosphate monohydrate]were synthesized at Shionogi Research Laboratories (Shionogi & Co., Ltd., Osaka, Japan) and were confirmed to have radiochemical purities above 99% by HPLC. The labeled positions of the test drugs are shown in Fig. 1. All other reagents were of analytical grade.

Perfusion Method.

Male Sprague-Dawley rats (Clea Japan, Inc., Tokyo) age 9 weeks were used. The rats were anesthetized with pentobarbital; the occipital artery, superior thyroid artery, and pterygopalatine artery were clotted or ligated; and the external carotid artery was cannulated in a retrograde manner with a polyethylene tube (PE50; Intramedic, Sparks, MD) (Takasato et al., 1984). Krebs-Henseleit buffer (118 mM NaCl, 14.7 mM KCl, 2.5 mM CaCl₂, 1.2 mM MgSO₄, 1.2 mM KH₂PO₄, 25 mM NaHCO₃, 10 mM d-glucose, pH 7.4) was used as a perfusate. This buffer was bubbled with a mixture of 95% O₂ and 5% CO₂, filtered through Millex HV filter (Millipore, Bedford, MA), incubated at 36°C, and infused with an infusion pump (Harvard Apparatus, S. Natick, MA). For the determination of residual perfusate in the brain, labeled compound and labeled sucrose were added to this perfusate. Immediately after perfusion, the animals were decapitated, and the forebrain was separated from the cerebellum and medulla oblongata. After removal of the pia mater and choroid plexus, the brain was dissolved with tissue solubilizer (Soluen 350; Packard, Meriden, CT), and radioactivity was determined with a liquid scintillation counter (Tri Carb 2200 CA; Packard).

Local Cerebral Perfusate Flow in Rat Forebrain.

A mixture of 7.4 kBq (1.03 μg)/ml [¹⁴C]diazepam and 37 kBq (0.017 μg)/ml [³H]sucrose was perfused for 10 s, and the animal was decapitated. Local cerebral perfusate flow (LCPF: represented Q in equation) was obtained using the equation:Equation 1where C_br(T), the parenchymal brain concentration (dpm/g) of [¹⁴C]diazepam at time T was calculated as the measured brain concentration minus intravascular tracer concentration, with the latter equal to the product of regional vascular volume measured by [³H]sucrose multiplied by perfusion fluid concentration of tracer.C_in is the concentration (dpm/g) of unbound [¹⁴C]diazepam in the perfusate, andT is perfusion time (s).

Apparent Exchangeable Fraction in Brain Microcirculation Determined with Perfusion Method.

Apparent exchangeable fractions for 3.7 kBq (0.36 μg)/ml [¹⁴C]caffeine, 3.7 kBq (1.3 μg)/ml [¹⁴C]S-312-d, and 37 kBq (18 μg)/ml [¹⁴C](+)-S-145 Na were measured with the perfusion method. Each ¹⁴C-labeled tracer and [³H]sucrose were added to perfusate. The BSA concentration in perfusate was varied between 0 and 1 mM. Perfusion with [¹⁴C]caffeine and [¹⁴C]S-312-d was performed for 10 s, and perfusion with [¹⁴C](+)-S-145 Na was performed for 30 s; the value ofC_br(T)/(C_in,tot· T · Q) was obtained for each concentration of BSA. C_br(T) of each tracer was the parenchymal brain concentration.K_d,app was calculated by fitting to eq. 2 (refer to for details):Equation 2The apparent exchangeable fraction (f_app) was obtained from eq. 14(Pardridge and Fierer, 1990) given in the .

Apparent Exchangeable Fraction in Brain Microcirculation Determined with i.c.a. Injection Method.

Each rat was anesthetized with pentobarbital, and 0.2 ml of a mixture of test compound and a reference compound was rapidly injected from the carotid artery. The animal was decapitated at 5 s after the injection, and the brain uptake index was calculated from the ratio of test to reference values of the administered solution and that in the brain (Oldendorf, 1970). The test compounds used were 3.7 kBq (0.04 μg)/ml [³H]diazepam, 37 kBq (3.6 μg)/ml [¹⁴C]caffeine, 74 kBq (0.034 μg)/ml [³H]propranolol, 18.5 kBq (6.4 μg)/ml [¹⁴C]S-312-d, 200 kBq(100 μg)/ml [¹⁴C](+)-S-145 Na, and 37 kBq (15 μg)/ml [¹⁴C]S-8510. [³H]Water or [¹⁴C]butanol was used as the reference compound. For comparison of the values obtained with different reference compounds, brain uptake index was converted to the extraction (E) with N-isopropyl-p-iodoamphetamine (Pardridge et al., 1985). The BSA concentration in the administered solution was varied between 0 to 1 mM. E was measured for each concentration, and the apparent dissociation constant (K_d,app ) was calculated by fitting to eq. 3, derived from the model in Fig. 2(Pardridge and Landaw, 1984):Equation 3where k₃ is the rate constant of drug transport from blood to brain (s⁻¹);t is the mean brain capillary transit time (s); and [P] is the concentration of BSA (M). The apparent exchangeable fraction (f_app) was obtained from eq. 14(Pardridge and Fierer, 1990) given in the . The fitting of eqs. 2 and 3 to brain permeability-protein concentration in the perfusate curve was performed with least-squares regression analysis (Yamaoka et al., 1981). The K_{d, app} ± S.D. predicted on the basis of fitting the experimental data to eqs. 2 and 3 is shown in Table 2. The data were analyzed by using the Damping-Gauss-Newton method, except for S-8510 (in the i.c.a. injection method) and caffeine (in the perfusion method), which analyzed with the Simplex method.

PS Products Determined with Perfusion Method.

PS products for 3.7 kBq (0.04 μg)/ml [³H]diazepam, 37 kBq (3.6 μg)/ml [¹⁴C]caffeine, 37 kBq (0.02 μg)/ml [³H]glucose, 3.7 kBq (1.3 μg)/ml [¹⁴C]S-312-d, 37 kBq (18 μg)/ml [¹⁴C](+)-S-145 Na, and 2.5 kBq (1 μg)/ml [¹⁴C]S-8510 were measured using eq. 4 (Takasato et al., 1984). Rats were perfused with buffer containing 0.5 mM BSA at the rate of 3.5 ml/min. Perfusion time for [¹⁴C]S-312-d was 10 s, and that for the other drugs was 30 s.C_br(T) of each tracer was the parenchymal brain concentration.C_in is the concentration of unbound tracer in the BSA-containing perfusate_. In the case of [³H]diazepam, however, the value within parentheses in eq. 4 was negative for half of the measurements, and PS could not be calculated. PS was therefore obtained using eq. 9 in the from k₃t, which was obtained at the time of the K_{d, app} calculation by using eq. 3.Equation 4

Protein Binding In Vitro.

Each drug was dissolved in the Krebs-Henseleit buffer containing 0.5 mM BSA (Fraction V; Sigma, St. Louis, MO), and protein binding was measured with the ultrafiltration method using an MPS-3 filter (Amicon, Beverly, MA). In this experiment, the adsorption of each drug to the PMS-3 filter was measured and used to correct the protein binding.

Results

LCPF in Rat Forebrain.

The flow rate of perfusate from pump was set at three levels (2.5, 3.5, and 5 ml/min), and LCPF was measured in BSA concentrations in perfusate from 0 to 1 mM (Table1). At each pump flow rate, LCPF in the hemisphere of the forebrain was smaller than the pump flow rate, indicating that perfusate introduced from the carotid artery was sent to several regions other than the forebrain. When the flow rate of perfusate containing 0.5 mM BSA was doubled from 2.5 to 5 ml/min, LCPF increased only about 52%. Thus, deviation of the perfusate outside the forebrain increased when the pump flow rate was elevated. LCPF was little affected by BSA concentrations up to 0.2 mM at any flow rate but gradually decreased at higher concentrations of BSA, probably due to increased vascular resistance resulting from increased viscosity.

Apparent Exchangeable Fraction in Brain Microcirculation Determined with Perfusion Method.

Decreases inC_br(T)/(C_in,tot· T · Q) caused by BSA concentration in the perfusate are shown in Fig. 3. The values obtained with the perfusion method are indicated by the symbols, and the values predicted on the basis of fitting the experimental data to eq. 2 are indicated by solid lines. The value ofC_br(T)/(C_in,tot· T · Q) for caffeine was not affected by BSA concentrations, that for (+)-S-145 Na sharply decreased, and that for S-312-d showed an intermediate value. TheK_d,app of caffeine calculated in this experiment was 19,999 mM, whereas that of (+)-S-145Na was 0.0212 mM (Table 2).

Apparent Exchangeable Fraction in Brain Microcirculation Determined with i.c.a. Injection Method.

The extraction (E) of each drug was determined with the i.c.a. injection method. The experimentally observed values are indicated by the symbols, and theE predicted by fitting the experimental data to eq. 3 is indicated by a solid line (Fig. 4). Although E decreased with increase in BSA concentration in the i.c.a. injection method, the degree of decrease varied depending on the drug. Extraction of S-8510 was not affected by BSA at all, whereasE of diazepam and caffeine were slightly affected and that of (+)-S-145 Na decreased sharply with increasing BSA concentration. The K_d,app of S-8510 was 223,121 mM, whereas that of (+)-S-145 Na was 0.0358 mM, indicating the slow dissociation of latter compound from BSA (Table 2).

The values of the brain permeability parameterC_br(T)/(C_in,tot· T · Q) obtained with the perfusion method and E obtained with the i.c.a. injection method for (+)-S-145Na are compared in Fig. 5. TheC_br(T)/(C_in,tot· T · Q) reflects the ratio of drug concentration in the brain at time T to the total amount of drug perfused in the brain until time T, whereas E compares the ratio of the reference compound to the test compound in the perfusate with the corresponding ratio in the brain. Although different methods were used, similar changes were noted for these indices of BBB permeability by BSA. These findings showed that both methods yield an accurate determination of BBB permeability.

The apparent dissociation constants (K_d,app ) and apparent exchangeable fraction (f_app) obtained with the i.c.a. injection method and the perfusion method in the BSA concentration (0.5 mM) are shown in Table 2. Unbound fractions measured with the ultrafiltration method in vitro are also shown for comparison.

The apparent exchangeable fraction was higher than the in vitro unbound fraction for every compound tested. These observations indicate that compounds bound to BSA under static conditions will dissociate under dynamic conditions in the cerebral circulation. The apparent exchangeable fraction of S-312-d was 38% and much higher than the in vitro findings of 1% in the unbound fraction, indicating that this compound has a high protein-binding ratio but readily dissociates in the cerebral circulation. The apparent exchangeable fraction of diazepam was about 8 times the in vitro unbound fraction, whereas those of other compounds were less than twice as high as the in vitro unbound fraction. For caffeine, there was little difference between the two. It has been reported that the in vitro unbound fraction of propranolol was 0.3 in buffer containing 4% BSA (Gariepy et al., 1990) and 0.29 in buffer containing 3% BSA (Jones et al., 1984). These values are almost the same as the apparent exchangeable fraction (f_app) in the cerebral circulation observed in the present study.

The K_d,app obtained with the perfusion method was slightly lower than that obtained with the i.c.a. injection method for S-312-d and (+)-S-145Na. TheK_d,app of caffeine was 4.99 mM with the i.c.a. injection method but very high with the perfusion method. Although the K_d,app obtained with the two methods appeared to differ, the apparent exchangeable fraction (f_app) at 0.5 mM BSA calculated from theK_d,app was 90.9% for the i.c.a. injection method and not significantly different from the value of 100% for the perfusion method. Indeed, theK_d,app can vary markedly based on the fitting estimation for drugs whose permeability is little affected by BSA, but the apparent exchangeable fraction, which is estimated fromK_d,app, is approximately equal to the almost unbound fraction.

Comparison of Unbound Fraction In Vitro and Apparent Exchangeable Fraction In Vivo of Diazepam.

The unbound fraction of diazepam was extremely decreased by 0.017 mM BSA in the buffer solution in vitro and decreased further as the BSA concentration increased (Table3). On the other hand, the apparent exchangeable fraction of diazepam was decreased only a little by an increase in the concentration of BSA. In 0.17 mM BSA, the ratio of the apparent exchangeable fraction to the unbound fraction was 3.6, and in 0.5 mM BSA, it was 8; this meant that an 8-fold higher concentration of unbound fraction of diazepam existed in the cerebral circulation and more drug contributed to permeation of BBB under the usual experimental conditions of measuring PS product used in the present study.

PS Products.

PS products calculated using the apparent exchangeable fractions measured with the i.c.a. injection method and the perfusion method are compared in Table4. Diazepam exhibited the highest BBB permeability, 21.3 × 10⁻³ ml/s/g, followed by caffeine, S-312-d, glucose, (+)-S-145Na, and S-8510, in this order. All of the compounds exhibited BBB permeability equal to or higher than that of glucose. Because glucose exhibited no protein binding in the in vitro ultrafiltration method, its unbound fraction was presumed to be 100% in the cerebral circulation in vivo. PS products calculated from the apparent exchangeable fractions obtained with the perfusion method were slightly high for S-312-d and (+)-S-145 Na.

PS products calculated from the in vitro unbound fractions measured with the ultrafiltration method were high for all of the compounds tested. In the case of diazepam and S-312-d, for which large differences between the apparent exchangeable fraction and the in vitro unbound fraction were observed, values within the parentheses in eq. 4became negative and calculations could not be performed using the in vitro unbound fraction. Thus, more dissociation from protein may occur in the cerebral circulation in vivo than under the static condition in vitro.

Discussion

In general, drugs penetrating into tissues are thought to be those in the unbound fraction in blood, and in vitro values are usually used to estimate drug transport. Detailed analytical methods are available for measuring protein binding through the use of equilibrium dialysis and ultrafiltration. Two protein-binding sites for diazepam and (+)-S-145 Na were indicated with Scatchard plot analysis and Michaelis-Menten plot analysis, and the corresponding dissociation constants, K_d1 andK_d2, of each compound were obtained (Table 2). Caffeine and S-8510 yielded straight lines on Scatchard plots and thus appeared to each have one binding site. In contrast, only the apparent dissociation constant K_d,appcan be obtained by fitting data obtained with the i.c.a. injection and perfusion methods. Although the values ofK_d1 and K_d,appwere almost the same for caffeine and (+)-S-145 Na, they differed greatly for diazepam and S-8510. The significance of these differences is unclear at present. K_d,app was measured for dissociation from albumin in the cerebral circulatory system assuming a single binding site. Although there may be sites with greater or lesser dissociability for compounds from albumin, theK_d,app was measured for overall dissociation from albumin in the cerebral circulation.

Pardridge and coworkers have reported the permeation of protein-bound compounds through the BBB (Cornford et al., 1983;Pardridge et al., 1983; Terasaki et al., 1986). In addition, albumin-bound drugs have been reported to be transported into the liver (Weisiger et al., 1981; Forker and Luxon, 1983).

In the present study, we demonstrated the participation of protein-bound drugs in BBB permeation for diazepam, caffeine, propranolol, S-312-d, (+)-S-145Na, and S-8510 with the i.c.a. injection method and for caffeine, S-312-d, and (+)-S-145 Na with a new analytical procedure of the perfusion method. Although the mechanism by which protein-bound drugs permeate tissues is not known, the report ofHorie et al. (1988) that conformational change of albumin occurs in contact with isolated rat hepatocytes is of interest because this would suggest that dissociation of compounds from protein may occur due to the conformational change of albumin when it comes into contact with endothelial cells in brain capillaries. If this does occur, the strength of the hydrogen bonding, which plays an important role in the binding of drugs to albumin, may be related to the apparent exchangeable fraction.

We developed a new method to measure the apparent exchangeable fraction from data obtained with the perfusion method and compared the values with those obtained with the i.c.a. injection method that had already been reported by Pardridge and Landaw (1984). TheK_d,app values for (+)-S-145 Na and S-312-d obtained with the perfusion method were nearly the same as those obtained with the i.c.a. injection method. On the other hand,K_d,app for caffeine obtained with the perfusion method was very large and differed greatly from that obtained with the i.c.a. injection method (Table 2). However, the apparent exchangeable fractions (f_app) of caffeine calculated from K_d,app were 100% for the perfusion method and 90.9% for the i.c.a. injection method and thus differed only a little. In the case of caffeine, which undergoes less dissociation from BSA and has a large apparent exchangeable fraction, the differentK_d,app may have been predictable on the basis of fitting estimation from different experimental methods.

The K_d,app obtained under anesthesia with ether was smaller than that obtained with pentobarbital because blood flow is more rapid under ether anesthesia (Pardridge and Fierer, 1990). If this is true, it should not be possible to apply the apparent exchange fraction obtained with the i.c.a. injection method to the perfusion method when LCPF is larger than the cerebral blood flow of the i.c.a. injection method. Therefore, theK_d,app for (+)-S-145 Na was measured by changing the perfusion rate (Fig. 6). The effects of BSA on theC_br(T)/(C_in,tot· T · Q) for (+)-S-145 Na were unchanged when the perfusion velocity was increased from 2.5 to 5 ml/min, and no difference in K_d,app was found. These observations confirmed that K_d,app was not affected by a change in the blood flow rate.

In conclusion, the following findings validated the use of the perfusion method for determination of the apparent exchangeable fraction: 1) the apparent exchangeable fractions obtained with the i.c.a. injection method reported by Pardridge and colleagues were equivalent to those obtained with our perfusion method, 2) the effect of BSA onC_br(T)/(C_in,tot· T · Q) measured with the perfusion method was similar to that on E measured with the i.c.a. injection method, and 3) equal and stable values ofK_d,app were obtained with different perfusion velocities.