Main

Kernicterus, which is characterized by yellow staining of the basal ganglia and of some subcortical nuclei, was first described more than a century ago(1, 2). For a number of years bilirubin entry into brain nuclei was thought to occur only as free (unbound) bilirubin(3–5). Indeed, during a single capillary transit, significant uptake of free, but not albumin-bound, bilirubin has been shown to occur(6). However, the dye can also be detected in the brain even when the molar concentration of bilirubin does not exceed that of albumin. In that case, most of the dye probably enters the brain after being dissociated from the albumin-bilirubin complex during passage through the brain vasculature(7). In this situation bilirubin entry into the brain correlates with the total serum bilirubin concentration. When the bilirubin:albumin ratio exceeds unity, bilirubin entry into the brain is more closely related to the concentration of free bilirubin in serum(7).

In autopsy specimens from kernicteric infants, the accumulation of bilirubin is observed mostly in the globus pallidus, subthalamic nucleus, hippocampus, substantia nigra, hypothalamus, cranial nerve nuclei (oculomotor, vestibular, cochlear and facial), brainstem reticular formation, and cerebellar dentate nucleus(8–11). Regional differences in brain bilirubin uptake have also been described in newborn piglets, both in normal conditions(12) or during hyperosmolar(13) or hypercarbic opening of the blood-brain barrier(14), with a higher uptake in cerebellum and brainstem compared with midbrain and cerebral cortex. However, this regional distribution pattern is almost lost by 2 wk of age in the piglet(12) and absent in the adolescent rat during hypercarbia, hyperosmolality, or sulfisoxazole displacement(15). Likewise, no regional differences were found in [3H]bilirubin binding to adult rat brain slices(16). However,[3H]bilirubin was shown to bind preferentially to neurons rather than glia(16, 17).

Because there are no data available on the regional uptake of bilirubin in the rat pup, in the present study, we measured the blood-to-brain transfer constant of bilirubin in rats at P10 and P21 to correlate the regional uptake rates for the dye to the corresponding regional metabolic changes previously recorded at the same age during hyperbilirubinemia(18). In the first part of the study, we measured the basal rate of[3H]bilirubin entry into the brain in the presence of trace quantities of the dye. In the second part of the study, the regional uptake of[3H]bilirubin was assessed after the induction of severe hyperbilirubinemia, as previously described(18). The ages of the rats in the present study were chosen to be representative of the human neonatal period. Indeed, the 10-d-old rat is considered to be the equivalent to a human newborn in terms of cerebral maturation, whereas the 21-d-old rat is close to a 9-mo-old infant(19, 20).

METHODS

Animals. Adult Sprague-Dawley rats (Iffa-Credo Breeding Laboratories, L'Arbresle, France), one male and two females by cage, were housed together in mating groups for 5 d and constantly maintained under standard laboratory conditions on a 12:12 h light/dark cycle (lights on at 0600 h). After delivery, litter sizes were reduced to 10 pups for homogeneity. The experiments were performed on naive or hyperbilirubinemic rats at P10 and P21 (day of birth was considered as d 0). All experiments were conducted in conformity with the Guide for Research Involving Animals and Human Beings.

For the measurement of cerebrovascular permeability with[3H]bilirubin, a femoral artery and vein were catheterized under light halothane anesthesia with PE 10 polyethylene catheters (Clay Adams, Parsippany, NJ; inside diameter 0.28 mm, outside diameter 0.61 mm) filled with 100 IU/mL heparin in isotonic saline. For the measurement of regional blood volume with [14C]sucrose, only the femoral vein was catheterized. Catheters were threaded under the skin, up to the neck, and led outside through a small opening in the skin. A loop was made with the end of each catheter, which was placed under the skin. The small opening in the skin was sutured. The animals were returned overnight to their mother in their normal environment.

Induction of hyperbilirubinemia. On the day of the experiment, the rats were transferred to a rectangular plexiglass box to prevent free ambulation of the perfused animals, especially at P21. A bilirubin solution(32 mg/mL) in 0.5 M NaOH, 0.055 M phosphate buffer, and 5% BSA at pH 7.4(10:20:70) was perfused through one femoral vein. The perfusion of the bilirubin solution (final concentration, 3.2 mg/mL or 5.5 mM in 8 mM BSA solution) was performed in two steps. First, a loading dose of 160 mg/kg was administered over 15 min. The speed of the infusion was then reduced to 64 mg/kg/h, and the perfusion was continued for 90 min leading to a total infusion time of 105 min. Control animals received an equivalent volume of PBS over the same period of time. Control animals did not receive albumin which was given in this protocol only to bind bilirubin molecules. Moreover, because of the high photosensitivity of bilirubin, the whole experiment was performed in low intensity red light. The measurement of blood-to-brain transfer constant for [3H]bilirubin was started at 90 min after the onset of bilirubin infusion and continued at the same speed over the 15-min experimental period.

Measurement of albumin and bilirubin. Brain bilirubin content was determined by chloroform extraction according to the method described by Bratlid and Winsnes(21).

The measurement of total plasma bilirubin was performed in the presence of caffeine according to the method originally described by Jendrassik and Grof(22). Plasma samples were mixed with distilled water, diazo reagent, and methanol. Bilirubin reacts with sulfanilic acid to form pink azoic pigments. Samples of 50 mg/L total bilirubin were used for standardization. Spectrophotometric readings were performed at 540 nm.

For the determination of albumin, plasma was mixed with albumin reagent containing bromcresol green according to the method of Doumas et al.(23). After 30 s, spectrophotometer readings were performed at 630 nm.

Measurement of cerebrovascular permeability. To determine the cerebrovascular permeability, control rats were injected over 20 s with mixed isomers of [vinyl-3H]bilirubin (3.16 MBq/kg, specific activity, 1.85 GBq/mmol, Amersham, Little Chalfont, Buckinghamshire, UK) in the femoral vein. In bilirubin-exposed rats, the tracer was administered at 90 min after the onset of cold bilirubin perfusion. Blood samples were drawn from the femoral artery at timed intervals over the 15-min experimental period. Blood samples were immediately centrifuged, and the plasma concentration of[3H]bilirubin was determined on 5-μL plasma samples in a scintillation counter (Beckman, model LS 1801).

At approximately 15 min after the pulse of [3H]bilirubin, the animals were killed by decapitation. Brains were rapidly removed and frozen in isopentane chilled to -30°C, coated with chilled embedding medium(carboxymethylcellulose 4% in water), and stored at -80°C in plastic bags before sectioning and autoradiography. The brains were then cut into 20-μm coronal sections at -22°C in a cryostat. Sections were picked up on glass coverslips and dried on a hot plate (60°C). Sections were autoradiographed for 8-14 wk on tritium Hyperfilm along with calibrated 3H standards(Amersham Corp., Arlington Heights, IL). Adjacent sections were fixed and stained with thionine for histologic identification of specific nuclei.

The autoradiographs were analyzed by quantitative densitometry with a computerized image processing system (Biocom 200, Les Ulis, France). OD measurements for each structure anatomically defined according to the developing rat brain atlas of Sherwood and Timiras(24) were made bilaterally in a minimum of four brain sections. All densitometry was conducted without knowledge of the treatment of the animal. Tissue3 H concentrations determined from the OD of the autoradiographic representations of the tissues were quantitated based on the calibration curve obtained from the standards.

Calculation. The blood-to-brain transfer constant(Ki) was calculated in control and bilirubin-exposed rat brains for [3H]bilirubin from the tissue and plasma radioactivity data assuming a two-compartment model and unidirectional transfer of the tracer and using the equation developed by Ohno et al.(25) and adapted to molecules with low permeabilities, as is the case for bilirubin(26, 27): Equation where Cbr represents the amount of tracer in the brain per unit mass of tissue (dpm/g) at time T,T is the duration of the experiment (min), Cpl is the arterial plasma concentration (dpm/mL), V represents the regional blood volume (mL/g), and Cwb is the tracer concentration in whole blood at the time of decapitation (dpm/mL).

However, there is some controversy regarding the rate of clearance of bilirubin from the brain which ranges from about 20 min for one study(28) to 1.7 h in another(29). The model used in the present study measured only the rate of entry of bilirubin into the brain. Therefore, if the half-life of bilirubin inside the brain compartment is as short as 20 min(28), the apparent rate of transfer of bilirubin into brain measured by means of the present technique would certainly slightly underestimate the actual rate of transfer of the anion from plasma to brain.

Measurement of regional cerebral blood volume. To determine the possible changes in the vascular compartment induced by hyperbilirubinemia and calculate the amount of tracer that crossed the capillaries unidirectionally during the experiment, the regional cerebral blood volume (V) was determined in saline- and bilirubin-exposed rats by means of[14C]sucrose (specific activity, 18.5-22.4 GBq/mmol, Du Pont de Nemours, Les Ulis, France). The radiotracer was injected at 90 min after the onset of bilirubin infusion as a 1- and 3-μCi bolus at P10 and P21, respectively. The regional cerebral blood volume was taken as the ratio dpm·g-1 of brain/dpm·g-1 of whole blood at the time of death, i.e. 2 min. Because sucrose crosses the blood-brain barrier at a very slow rate, it can be considered as restricted to the cerebral vessel compartment during the first 2 min of the experiment unless the blood-brain barrier is disrupted. Rats were killed by decapitation at 2 min after tracer injection, and body blood samples were collected. The brain was dissected into nine regions of interest, i.e. cerebral cortex, hippocampus, striatum, hypothalamus, thalamus, inferior and superior colliculi, brainstem, and cerebellum. Blood and brain samples were transferred to preweighed scintillation vials that were immediately covered and weighed after blood and tissue collection. The blood samples were then treated with 0.5 mL of a mixture of tissue solubilizer, Optisolv (FSA Laboratory Supplies) and isopropanol (1:2, v:v) followed by 0.5 mL of hydrogen peroxide (30%). Tissue samples were digested by 1 mL of pure Optisolv. Blood and tissue concentration of [14C]sucrose were then determined by liquid scintillation counting.

Physiologic variables. Just before the onset of cold bilirubin infusion and before isotope injection, mean arterial blood pressure was measured with an air-damped mercury manometer. Hematocrit was determined from blood samples collected into capillary tubes at about 10 min after isotope injection. Just before the onset of cold bilirubin infusion and before[3H]bilirubin injection, a sample of 40-80 μL of arterial blood was taken for measurement of blood pH, Po2 and Pco2, which were measured with a blood gas analyzer (Corning, model 158, Corning Medical and Scientific, Halstead, England). P10 rats were maintained on a heating pad to keep their body temperature in the physiologic range.

Statistical analysis. The values of cerebral blood volume, blood-to-brain transfer constant for [3H]bilirubin, and physiologic variables underwent a two-way analysis of variance. Thereafter, values in the control P10 or P21 groups were compared with those in the age-paired bilirubin-infused group by means of a t test for independent groups.

RESULTS

Effects of Hyperbilirubinemia on Physiologic Variables

In the rats receiving only the bolus of tracer amounts of[3H]bilirubin, physiologic variables were in the normal range and comparable to our previously published data(18, 30). When exposed to a high dose of cold bilirubin, P10 rats showed a significant decrease in arterial pH and arterial blood pressure at 90 min of perfusion (Table 1). At P21, the above variables were not affected by bilirubin perfusion. The lack of difference between the hematocrits in control and bilirubin-exposed rats shows that the bilirubin-albumin perfusion did not induce significant hemodilution.

Table 1 Effects of hyperbilirubinemia on physiologic variables in control (C) and bilirubin-intoxicated (Bi) immature rats
Full size table

In perfused rats, plasma bilirubin and albumin levels were similar at both ages, reaching 200-230 and 360-397 nmol/mL, respectively. Brain bilirubin concentration was 40% lower after a 90-min perfusion in P21 compared with the P10 rats (Table 1). Plasma albumin concentration was increased by 34 and 64% in perfused rats at P10 and P21, respectively. Albumin:bilirubin ratios reached 1.7-1.8 in perfused rats at both ages. In control rats not receiving any albumin and only tracer amounts of[3H]bilirubin, the ratio of albumin:bilirubin calculated from the measured albumin concentration at zero time and the concentration of[3H]bilirubin present in blood after the injection of the tracer ranged from 15 to 16 at both ages.

Blood-to-Brain Transfer Constant for Bilirubin

Analysis of variance. The effect of age was highly significant in all brain regions reaching a p value of 1-3 × 10-4. The effect of treatment was also highly significant in most brain areas(10-4 < p < 2 × 10-3), except in ventromedian hypothalamus (p = 0.035) and in the cochlear and lateral geniculate nuclei where the effect of treatment was not significant. There was also a significant interaction between age and treatment in most brain areas studied with a highly significant p value of 5 × 10-3 to 10-4 in most brain regions, except in the cochlear nucleus, lateral lemniscus, ventromedian hypothalamus, and mediodorsal thalamus where there was no significant interaction.

Effect of age on the regional blood-to-brain transfer constant for bilirubin. At P10, blood-to-brain transfer constant (Ki) for bilirubin reached values ranging from 0.07 to 0.12 μL/g/min, except in the auditory nerve (Table 2), dentate nucleus (seeTable 4), and hypothalamic and thalamic areas (seeTable 5), where it ranged from 0.41 and 0.47μL/g/min.

Table 2 Blood-to-brain transfer constant for bilirubin in sensory systems of control (C) and bilirubin-intoxicated (Bi) immature rats
Full size table
Table 4 Blood-to-brain transfer constant for bilirubin in motor and white matter areas of control (C) and bilirubin-intoxicated (Bi) immature rats
Full size table
Table 5 Blood-to-brain transfer constant for bilirubin in hypothalamus and thalamus of control (C) and bilirubin-intoxicated (Bi) immature rats
Full size table

At P21, Ki for bilirubin was reduced compared with P10, with values ranging from 0.03 to 0.06 μL/g/min in most brain areas. The value of bilirubin brain uptake was not significantly different at P21 from P10 only in 6 areas of the 39 studied. The areas were the auditory and olfactory cortices, the inferior colliculus, the superior olive(Table 2), the prefrontal cortex, and the nucleus accumbens (Table 3).

Table 3 Blood-to-brain transfer constant for bilirubin in limbic and functionally nonspecific areas of control (C) and bilirubin-intoxicated (Bi) immature rats
Full size table

During hyperbilirubinemia, Ki for bilirubin was significantly lower at P21 than at P10 in all brain areas studied(Tables 2–5).

Effects of hyperbilirubinemia on the blood-to-brain transfer constant for bilirubin. Hyperbilirubinemia induced large increases in the blood-to-brain transfer constant for bilirubin which were more pronounced at P10 than at P21 (Tables 2–5). At P10, increases were not significant in three regions, i.e. lateral lemniscus, cochlear nucleus (Table 2), and ventromedian hypothalamus (see Table 5). In most other regions, bilirubin uptake was increased to over 200% of control with very large increases (over 450%) in the superior olive and the olfactory cortex(Table 2) as well as in most limbic areas(Table 3), posterior hypothalamus, and thalamus(Table 5). High increases (over 350% of control) were also seen in the auditory and visual cortex, lateral geniculate body(Table 2), ventral tegmental area, pontine gray, substantia nigra pars compacta, and genu of the corpus callosum(Table 4).

At P21, hyperbilirubinemia induced increases in blood-to-brain transfer for bilirubin that were statistically significant in 16 brain areas, although their magnitude was lesser than in the P10 rat, mostly ranging from 50 to 200% of control values. Highest increases (over 200% of control) were located in the hippocampus (Table 3), sensory-motor cortex(Table 4), and anterior and posterior thalamic nuclei(Table 5). More moderate increases (over 120% of control) were recorded in the visual and frontoparietal cortex, lateral geniculate body(Table 2), nucleus accumbens, ventral tegmental area, nucleus of the solitary tract (Table 3), and both parts of the substantia nigra (Table 4).

Cerebral Blood Volume

Cerebral blood volume was unchanged by hyperbilirubinemia in most regions and significantly increased at both ages (26-51%) in the superior and inferior colliculi and in the cerebellum (Table 6). Cerebral blood volume was decreased by bilirubin exposure in the hippocampus in P10 and P21 rats and in the thalamus in P21 rats.

Table 6 Regional cerebral blood volume of control (C) and bilirubin-intoxicated (Bi) immature rats
Full size table

DISCUSSION

The results of the present study represent the first regional approach of bilirubin entry into the brain of immature rats. They demonstrate that the blood-brain permeability to bilirubin is about twice as high in control rats at P10 than at P21. It is also relatively homogeneous throughout the brain, except in a few areas. However, after induction of hyperbilirubinemia, the blood-to-brain transfer constant for bilirubin is significantly increased, especially at P10.

In the basal case studied in the present report, bilirubin was injected into the animals in tracer amounts, and therefore all of the bilirubin present in the blood could be considered bound to albumin because the calculated ratio of albumin:bilirubin was close to 15. Under these conditions, a significant uptake of bilirubin was measured which confirms the data of Robinson and Rapoport(7) showing that even when the albumin:bilirubin ratio is high and most of the bilirubin is bound to albumin, bilirubin is still able to enter the brain in significant amounts, certainly after being dissociated from the bilirubin-albumin complex during the passage through the brain. This basal level of bilirubin uptake was quite homogeneous throughout the brain with only a few regions exhibiting a higher rate of bilirubin uptake, mostly in the auditory system in the P10 rat. In the P21 rat, there was no regional difference in the basal rate of blood-to-brain bilirubin transfer. The blood-to-brain transfer constant for bilirubin decreased about 2-3-fold in most regions and even 10-20-fold in areas such as the lateral lemniscus and cochlear nucleus compared with that of the P10 rat. The overall decrease in the cerebral permeability to bilirubin is in good accordance with that recorded for other molecules of variable molecular weight. For example the transfer of inulin, albumin, and sucrose from blood to brain decreases about 40-60% between P2 and P16-P20(31–33). Likewise, brain permeability to mannitol and urea decreases significantly between birth and 5 wk of postnatal age(34–36).

In the hyperbilirubinemic conditions designed in the present study, most of the bilirubin is still bound to albumin, because the albumin:bilirubin ratio reaches 1.7-1.8. In those conditions, the uptake of bilirubin was highly enhanced compared with the basal situation. This increased cerebral permeability to bilirubin might be partly explained by the 9-fold decrease in the albumin:bilirubin ratio between the basal situation and the hyperbilirubinemic conditions of the present study. However, the transfer of bilirubin to brain in hyperbilirubinemic conditions was not only increased but became also regionally more heterogeneous, especially in the P10 rat. It was more homogeneous in the P21 rat, as already reported in the piglet at 2 wk of postnatal age(12). In the present study, hyperbilirubinemia increased the blood-to-brain transfer constant for bilirubin 2-4-fold in most brain areas at P10. The highest rates of bilirubin uptake by the brain (over 0.70 μL/g/min) were observed in the hippocampus, posterior hypothalamus, thalamus, dorsal raphe, and auditory nerve. High rates of brain permeability to bilirubin were also recorded in many auditory areas, vestibular nucleus, globus pallidus, cerebellar cortex and nuclei, and several cortical regions. These high values of blood-to-brain transfer constant for bilirubin are observed in the absence of any disruption of the blood-brain barrier, as previously shown(18) but with a decreased albumin:bilirubin ratio. These data are in good agreement with the regional accumulation of bilirubin observed after autopsy in kernicteric infants showing that bilirubin is mostly stained in the globus pallidus, subthalamic nucleus, hippocampus, substantia nigra, hypothalamus, cranial nerve nuclei(oculomotor, vestibular, cochlear and facial), brainstem reticular formation, and cerebellar dentate nucleus(8–11). There are many parallels in the regional distribution of brain bilirubin staining between full-term infants with marked hyperbilirubinemia, premature infants without marked hyperbilirubinemia, Gunn rats(11), and the increased level of brain permeability to bilirubin observed in P10 hyperbilirubinemic rats in the present study. In the 2-d-old piglet, bilirubin has been shown to accumulate to a greater extent in the cerebellum and the brainstem than in the midbrain and cerebral cortex(12), but more precise regional data are missing in the species. However, conversely to in vivo observations in infants(11) and in vitro studies in rat brain slices or cell cultures(16, 17, 37), we did not notice any difference in the rate of uptake of bilirubin by the different cell layers of the hippocampus and the cerebellum. At P21, the regional rate of bilirubin uptake by the brain of hyperbilirubinemic rats was much less heterogeneous than at P10, being elevated only in the hippocampus and in some cortical regions.

The enhanced permeability recorded in the present study in P10 hyperbilirubinemic rats in the caudate nucleus and cerebral cortex may be related to the specific vulnerability of these structures to bilirubin. However, although the dye has been shown to accumulate in the caudate nucleus in kernicteric infants(8–11) and to induce motor deficits(38, 39), bilirubin staining has not been usually observed in the cerebral cortex of kernicteric infants, although bilirubin encephalopathy is known to lead to cognitive deficits(40, 41). The increased permeability to bilirubin in the caudate nucleus and the cerebral cortex during hyperbilirubinemia reported in the present study is also in good accordance with the specific metabolic pattern induced by severe hyperbilirubinemia at P10. Indeed, in these conditions, levels of metabolic activity are heterogeneous within these structures with an alternation of dark and light columns in the cerebral cortex and of dark and light dots in the caudate nucleus(18). The origin of this peculiar distribution of energy metabolism levels is unknown. One hypothesis is that it might be correlated to the distribution of cerebral blood vessels, as previously shown in hypoxic conditions(42, 43). Moreover, the particular vulnerability of the caudate nucleus to bilirubin may be related to its rich blood supply during brain development(13). Indeed, vessel proliferation in the developing brain is characterized by junctional leakiness and high nonspecific permeability(44, 45), which could lead to more entry of bilirubin into regions specifically rich in blood vessels in the immature brain. Bilirubin entry into the brain is also increased during regional hyperemia induced by hypercapnia in the piglet(14, 46), but becomes independent of blood flow rates after opening of the blood-brain barrier with hyperosmolar urea(13). We have not measured local rates of cerebral blood flow during hyperbilirubinemia. Rates of cerebral blood flow could be influenced by the infusion of the bilirubin solution. However, rates of cerebral blood flow have been shown to be coupled to cerebral metabolism in the P10 rat rat both in normal conditions(47), as well as during hypoxia(48) or seizures(49). Therefore, it is likely that, in the hyperbilirubinemic conditions studied in the present report, rates of local cerebral blood flow are not increased but rather decreased in the face of reduced rates of cerebral glucose utilization induced by hyperbilirubinemia(18). Thus, the increase in the blood-to-brain transfer constant for bilirubin is certainly not related to an increase in cerebral blood flow consecutive to the bilirubin perfusion. Moreover, rates of cerebral blood flow are quite homogeneous throughout the brain of the P10 rat, except in some posterior areas(47). Thus, there is no direct correlation between the regions with highest blood flow and those with increased bilirubin permeability. Indeed, in the present study, brain permeability to bilirubin is high in some posterior regions, although being low in others and also high in some anterior regions where basal levels of blood flow are low. The differences in regional blood-to-brain transfer constant for bilirubin are thus likely to be related to interregional differences in the proliferation of blood vessels associated with junctional leakiness and high nonspecific permeability(44, 45).

Moreover, the overall highly increased permeability of brain regions to bilirubin induced by hyperbilirubinemia in P10 compared with P21 rats may be partly related to hyperbilirubinemia-induced acidosis that alters the binding constant of the albumin-bilirubin complex, according to some authors(50, 51). Others, however, have failed to demonstrate such a change in albumin-bilirubin binding(52, 53). Acidosis increases the uptake of bilirubin by the brain of newborn guinea pigs(54) and produces gross evidence of kernicterus in newborn puppies(55). Moreover, in newborn infants, increased binding of bilirubin to albumin is achieved when acidosis is corrected(56). More recently, Bratlid et al.(57) showed that only respiratory acidosis augments bilirubin concentration in the rat brain, whereas metabolic acidosis, as is the case in the present study does not. In view of the confusing results in the literature, experiments run under conditions of corrected blood gases are necessary to demonstrate that the increased bilirubin uptake in the P10 rat is not related to acidosis. However, the small blood volume of rat pups at P10 makes it difficult to perform repeated blood sampling for monitoring of blood gases.

In conclusion, the present study shows that the very immature rat brain(P10) is very permeable to bilirubin with very high rates of uptake in the auditory regions and that brain permeability to bilirubin decreases with postnatal maturation. In addition, when exposed to high levels of bilirubin, the brain of the 10-d-old rat becomes very permeable to bilirubin and there is a good correlation between the regions most permeable to bilirubin and the staining of brain regions in infants with kernicterus. However, because the albumin:bilirubin ratio significantly decreases from about 15-16 in the control conditions to 1.7-1.8 in hyperbilirubinemia, more experimental data are still necessary to determine whether the changes in the blood-to-brain transfer constant for bilirubin are related to the decrease of this ratio and/or to the preexposure to hyperbilirubinemia.