Main

The pathogenesis of hyperbilirubinemic encephalopathy remains incompletely understood but central to its development is the passage of bilirubin across the BBB into the CNS. It is generally held that bilirubin can enter the brain when either 1) bilirubin not bound to albumin (free bilirubin) crosses the BBB by passive diffusion (1) or 2) the BBB is disrupted (1–3). Bilirubin, despite its high affinity for membrane lipids (4), normally demonstrates a low accumulation in the brain that may not be fully accounted for by protein binding in plasma, movement into cerebrospinal fluid (3), or intracerebral metabolism by bilirubin oxidase (5,6). In this respect, bilirubin is similar to many xenobiotics with high lipophilicity that share an unexpectedly low accumulation in the CNS (7,8). Many of these xenobiotics are substrates for an ATP-dependent integral plasma membrane transporter, P-gp (9,10).

P-gp is constitutively expressed in many tissues, including the capillary endothelial cells (7,11–16) and astrocytes (17) of the BBB. P-gp is a member of the ATP-binding cassette (ABC) superfamily and is encoded for by a family of genes referred to as the mdr (multidrug resistance) genes (18,19). The term multidrug resistance was chosen for these genes because P-gp is commonly expressed in tumors where it functions as a multidrug pump and is associated with the resistance of tumor cells to a wide variety of chemotherapeutic agents. The intronexon structure of the human mdr1 and mouse mdr1a genes are virtually identical (10,14), and this gene 1) encodes the major isoform of P-gp expressed in brain capillary endothelial cells (13,14), and 2) confers a multidrug resistance phenotype in humans (9,10). Several studies have demonstrated that P-gp limits the CNS influx and retention of a wide variety of unrelated lipophilic compounds, i.e. P-gp contributes to the barrier function of the BBB (14,20,21) and has broad substrate specificity (9,10).

Recently, Schinkel et al. (21) have generated a mdr1 a null mutant transgenic mouse line [mdr1a(-/-)] that does not express P-gp and shown that in the absence of BBB P-gp expression there is enhanced brain uptake of the exogenous neurotoxin ivermectin and lipophilic xenobiotics (14,20,21). Bilirubin is an endogenously generated lipophilic product of heme catabolism that may be neurotoxic when serum levels become markedly elevated. In the current study, we postulated that bilirubin is a substrate for P-gp and tested the hypothesis that brain bilirubin content after an i.v. bilirubin infusion would be increased in P-gp deficient mdr1a null mutant transgenic mice [mdr1a(-/-)] compared with P-gp-sufficient control animals.

METHODS

Studies were conducted on adult mdr1a(-/-) P-gp-deficient transgenic null mutant mice and age-matched mdr1a(+/+) P-gp-sufficient FVB strain controls supplied by Taconic (Germantown, NY). The details regarding the genesis of the mdr1a(-/-) transgenic line have been previously described by Schinkel et al. (14,20,21). The mdr1a P-gp isoform is expressed in abundance in brain capillary endothelial cells of FVB mice but is absent in the mdr1a(-/-) null mutant line (14,20,21). The experiment was approved by the Magee-Womens Hospital and Research Institute Institutional Animal Care and Use Committee.

Eighteen animals from each study group were lightly anesthetized with pentobarbital sodium (40 mg/kg i.p.), and a 27-gauge injection needle (Abbott, Chicago) was inserted into the tail vein. Bilirubin, 50 mg/kg, was infused i.v. over 5 min, and the injection needle was removed. The bilirubin solution for infusion was prepared as follows. Bilirubin (B4126; Sigma Chemical Co.) was dissolved in 0.1 N NaOH, stabilized with BSA (A7906; Sigma Chemical Co.) at a bilirubin: albumin molar ratio of 14, and diluted with Krebs-Ringer buffer at pH 7.4 to a concentration of 3 mg/mL (5,22). All bilirubin-containing vials/syringes were wrapped in tinfoil to reduce the photodecomposition of bilirubin.

Ten minutes (n = 9, each group) or 60 min (n = 9, each group) after completion of the i.v. bilirubin infusion the mouse was killed with 100 mg/kg pentobarbital i.p., the chest was opened, and a blood sample for total serum bilirubin was taken from the left ventricle. A catheter was placed in the ascending aorta, the descending aorta was ligated, and the brain was flushed in situ with isotonic saline (15 mL) to clear the brain vasculature of bilirubin (23). The brain was then removed intact, stripped of its meningeal covering, weighed, and immediately frozen in liquid nitrogen.

The brain bilirubin concentration was determined by acid chloroform extraction followed by diazotization (24). Total serum bilirubin was measured by the diazo method (25). The rate of clearance of bilirubin from the brain was estimated using the program Expofit (26). This program fits values of concentration by least squares to values of time according to the equation C = C0e-kT, and calculated the initial concentration, C0, the rate constant, k, and half-life. The program is an alternative to semilogarithmic plotting and assumes first order kinetics, a condition recently shown to be applicable to bilirubin clearance from the CNS (5).

Statistical methods (Minitab Statistical Software, PC Version Release 8.0, Minitab Inc., State College, PA) (27) included the Mann-Whitney rank procedure to compare mdr1a(-/-) and control brain bilirubin content, total serum bilirubin levels, and brain bilirubin half-lives. Statistical significance was established at p < 0.05. Data are reported as a mean ± SEM.

RESULTS

Animal characteristics as a function of study time period and group are presented in Table 1. Animal age, body weight, and brain weight were not significantly different between study time periods for a given group, or between study groups at a given time period.

Table 1 Study group characteristics*
Full size table

Total serum bilirubin levels as a function of study time period and group are shown in Figure 1. Mean total serum bilirubin levels approximated 190 µmol/L (11 mg/dL) 10 min after infusion and fell significantly to approximately 120 µmol/L (7 mg/dL) by 60 min postinfusion in both P-gp-deficient mdr1a(-/-) null mutant and wild type controls. No significant differences were noted in total serum bilirubin levels between groups at a given time period.

Figure 1
figure 1

Serum bilirubin levels (µmol/L) as a function of study time period and study group. P-gp-deficient mdr1a(-/-) null mutant mice are shown in solid bars and wild type (+/+) mice are shown in hatched bars. Data are reported as a mean ± SEM.

Full size image

Brain bilirubin content as a function of study time period and group are shown in Figure 2. Brain bilirubin content was significantly higher in P-gp-deficient mdr1a(-/-) null mutant mice as compared with wild type control animals 10 min after infusion (p < 0.05). Brain bilirubin content declined in both groups at the 60-min time point but remained significantly higher in P-gp-deficient mdr1a(-/-) null mutant mice as compared with wild type controls (p < 0.05). Brain bilirubin clearance, however, did not differ between groups with brain bilirubin half-lives (mean ± SEM) approximating 50-60 min in both mdr1a(-/-) null mutant P-gp deficient (61 ± 24 min) and wild type control (52 ± 16 min) animals (p = 0.76).

Figure 2
figure 2

Brain bilirubin content (nmol/g) as a function of study time period and study group. P-gp-deficient mdr1a(-/-) null mutant mice are shown in solid bars and wild type (+/+) mice are shown in hatched bars. *p < 0.05 mdr1a(-/-) vs wild type (+/+). Data are reported as a mean ± SEM.

Full size image

DISCUSSION

The principal finding of this study is that P-gp-deficient mdr1a(-/-) null mutant mice have significantly higher (∼2-fold) brain bilirubin content after an i.v. bilirubin load as compared with P-gp sufficient control animals. The similar brain bilirubin clearance rates from mdr1a(-/-) null mutant and control mice suggest that the increased brain bilirubin content of P-gp-deficient mdr1a(-/-) mice was due to enhanced brain bilirubin influx and not altered CNS bilirubin retention. These data provide indirect evidence that mdr1a P-gp may play a role in limiting the influx of bilirubin into the CNS.

P-gp is expressed in several normal tissues including particularly high expression in 1) brain capillary endothelial cells and astrocytes of the BBB (11–17), 2) epithelial cells lining the gastrointestinal tract and renal tubules (15); 3) the adrenal cortex (15); and 4) testes (11,15,16). In brain and testes, P-gp is postulated to serve a blood-tissue barrier function, offering protection against toxic metabolites and xenobiotics (11,16). Schinkel et al. (14,20,21) have recently demonstrated that P-gp can limit both the CNS influx and retention of numerous lipophilic compounds. In P-gp-deficient mdr1a(-/-) null mutant mice, 1) brain accumulation of the centrally neurotoxic pesticide ivermectin (14) and the carcinostatic drug vinblastine (14) increased by two orders of magnitude and 2) the brain penetration of several other compounds normally transported by P-gp substantially increased (28). Thus it appears that P-gp contributes to the barrier function of the BBB by excluding potentially neurotoxic compounds or extruding them from the CNS before the advent of cytotoxic effects.

To date, only a few endogenously generated compounds have been identified as substrates for P-gp. These include the steroid hormones cortisol, corticosterone, and aldosterone (29). Others have reported indirect evidence that bilirubin may be a substrate for P-gp (7,30). Such evidence includes 1) the ability of bilirubin to inhibit the photoaffinity labeling of P-gp by [125I]arylazidoprazosin (a photoaffinity probe for P-gp) (7), and 2) preliminary studies that demonstrate limited uptake of [3H]bilirubin by human multidrug-resistant variant cells compared with parent cells that do not express the mdr1 gene (30). In the current study, we postulated that bilirubin was a substrate for P-gp. We addressed this question using a unique P-gp deficient mdr1a(-/-) null mutant transgenic mouse line as generated by Schinkel et al. (14,20,21) at the Netherlands Cancer Institute. This model was chosen because mdr1a(-/-) null mutant mice are a well characterized transgenic line which do not express P-gp in the BBB as confirmed using both immunohistochemical and Western immunoblotting techniques (14,20,21). Although care is needed in interpreting results from the mdr1a(-/-) knockout because of possible compensatory mechanisms that may mask the detection of P-gp functions, such adaptations have not been identified in the mdr1a(-/-) BBB (31,32). Consistent with this observation is the finding that drugs effectively transported by P-gp in vitro were also noted to accumulate in the brains of mdr1a(-/-) mice (14,21,28).

Brain bilirubin content was almost 2-fold higher in mdr1a(-/-) null mutant animals as compared with wild type (+/+) controls. This difference in brain bilirubin content between control and mdr1a(-/-) lines was noted despite similar serum bilirubin levels and rates of brain bilirubin clearance across study groups. The latter suggests that P-gp did not alter CNS bilirubin retention once bilirubin had crossed the BBB. The higher brain bilirubin content of mdr1a(-/-) mice appears to be mediated by enhanced brain bilirubin influx. The latter is consistent with the "hydrophobic vacuum cleaner" model for P-gp proposed by Gottesman and colleagues (10,33,34) in which P-gp intercepts its substrates within the plasma membrane before they enter the cytoplasm and pumps them into the extracellular space. The increase in mdr1a(-/-) brain bilirubin was modest compared with that observed for some P-gp substrates, e.g. digoxin (21,31), cyclosporin A (21), and quinidine (35), but was nevertheless significant and in terms of CNS bilirubin entry on the order of that observed with respiratory acidosis (36), a condition hypothesized to enhance the risk for developing hyperbilirubinemic encephalopathy. The results of the current study suggest that intensified research on the role of P-gp in modulating brain bilirubin uptake is warranted.