Main

Plasma levels of unconjugated bilirubin (UCB) are elevated in almost all newborn infants. In some infants with markedly elevated plasma UCB levels, bilirubin causes neurotoxicity, sometimes resulting in permanent neurologic dysfunction (1). Management guidelines for jaundiced term and near-term infants, published by the American Academy of Pediatrics (AAP), are based on the premise that total serum bilirubin concentration (BT) is the best available predictor of risk for bilirubin-induced neurologic damage (BIND) (2). Clinical evidence has indicated, however, that BT, beyond a threshold value of 20 mg/dL, is a poor discriminator of individual risk for BIND (3,4). Since over 99.9% of total plasma UCB (BT) is bound to albumin or apolipoprotein D (5) and only unbound bilirubin can enter the brain across an intact blood-brain barrier, the level of unbound “free” bilirubin (Bf) should theoretically provide a more accurate indication of the risk of kernicterus.

Published studies of UCB toxicity in cell cultures, conducted with different types of albumin at varied molar ratios of bilirubin/albumin (6–9) reported an increase in cell damage depending on BT. The hypothesis that cell injury is correlated better with Bf was first suggested by Nelson et al. (10) in an in vitro study. Moreover, Ostrow et al. (11) reported a meta analysis of in vitro studies that suggested cytotoxicity was better correlated with Bf (11). This conclusion was however based on Bf calculations using published binding constants, rather than direct measurement in the culture medium, and was thus limited by unknown variations in binding associated with differences in the composition of the incubation media.

In jaundiced newborns, plasma Bf levels at any given BT or BT/albumin ratio can vary widely due to varying concentrations of albumin and apolipoprotein D, differences in the binding affinity, and/or the presence of inhibitors of binding (12). Bf, therefore, cannot be accurately predicted from the concentrations of BT and albumin in plasma or culture medium, and must be measured directly. A modified, enzymatic peroxidase method has been developed to measure Bf in plasma (12) and tissue culture media (13), with minimal dilution of the sample. This is important, since the binding affinity for UCB decreases markedly with increasing albumin concentration (12,13). This assay is based on the observation that, in the presence of peroxidase and peroxide, unbound bilirubin is oxidized to colorless compounds, whereas protein-bound bilirubin is protected from oxidation (14).

The present study, for the first time, directly tests the hypothesis that Bf, measured with the peroxidase method, rather than BT predicts the toxicity of UCB in several cell lines under a wide variety of incubation conditions, using 3(4,5-dimethiltiazolil-2)-2,5 diphenyl tetrazolium (MTT) reduction to assess cell viability.

MATERIALS AND METHODS

Materials.

UCB, DMSO (DMSO, HPLC grade), horseradish peroxidase (HRP) (EC.1.11.1.7/P 8125, Type I), hydrogen peroxide (H2O2, 30% wt/vol), fatty acid free human serum albumin (HSA), and fatty acid free BSA fraction V (BSA), sulfadimethoxine, Nutrient Mixture F-12 HAM (F-12), Eagle's Minimum Essential Medium (EMEM), 3(4,5-dimethiltiazolil-2)-2,5 diphenyl tetrazolium (MTT) and nonessential amino acid solution (MEM), were purchased from Sigma Chemical Co.-Aldrich, Milan, Italy. PBS, Dulbecco's modified Eagle's medium-high glucose (DMEMHG), penicillin, and streptomycin were purchased from Euroclone, Milan, Italy. FCS (FCS) and GlutaMAX™, obtained from Invitrogen (Carlsbad, CA), contained 24 g/L albumin. Chloroform, HPLC grade was obtained from Carlo Erba, Milan, Italy.

Cell culture.

A human neuroblastoma cell line (SHSY5Y) was grown under 5% CO2 in F-12/EMEM (1:1) supplemented with penicillin, streptomycin, GlutaMAX™, MEM and 15% FCS. HeLa, primary mouse embryo fibroblast (MEF) and Striatal precursor 2a1 cell lines (15) were grown in DMEMHG with penicillin, streptomycin and 10% FCS at 37°C, except for 2a1 cells that were incubated at 33°C under 5% CO2. SHSY5Y and Striatal precursor 2a1 cell lines were kindly provided by Dr S. Gustincich (SISSA, Trieste, Italy). MEF were prepared as described by Hertzog (16). The animal protocol was approved by the Animal Care and Use Committee of the University of Trieste.

Preparation of bilirubin/albumin systems.

UCB was purified by the method of Ostrow and Mukerjee (17), divided into 200 or 400 μg aliquots, and stored at −20°C until used. Just before use, an aliquot was dissolved in 66 or 133 μL DMSO, yielding a stock solution containing 5 mM UCB, which was added to the albumin-containing culture medium to attain the desired BT with a final DMSO concentration that did not exceed 1% (vol/vol).

Cell viability by MTT reduction.

Cells were plated in 24 multiwell plates, achieving 70–80% confluence by the following day. The growing medium was then discarded, cells were washed with PBS at 37°C and exposed to UCB dissolved in the respective culture medium containing either FCS 15% (54 μM albumin), 30 μM or 60 μM HSA, or 30 μM BSA. Control cells were incubated in the same medium to which had been added the same volume of DMSO as the added stock UCB solution. After incubation with the UCB, the medium containing UCB was removed and replaced with culture medium containing MTT 0.5 mg/mL (18). After incubating the cells with MTT for 2 h at 37°C, the medium was discarded and MTT formazan crystals dissolved by adding 0.4 mL isopropanol/HCl 0.04 M and gentle shaking for 2 h at 37°C. After centrifugation, absorbance values at 570 nm were determined in a LD 400C Luminescence Detector, Beckman Coulter, Milan, Italy. Results were expressed as percentage of MTT reduction by cells not exposed to UCB, which was considered as 100% viability.

Bf measurements.

The unbound bilirubin (Bf) concentrations were determined using a modification of the horseradish peroxidase assay (12) as described by Roca-Burgos et al. (13). The method involves minimal dilution of the sample, minimizing the effect of dilution of the albumin concentration on binding affinity (19).

Effect of different albumin preparations on Bf levels and time course of toxicity.

The relationship of Bf, BT, and different albumin preparations on binding and toxicity were examined by incubating cells in media containing 30 or 60 μM HSA, 30 μM BSA or 10% FCS and varying bilirubin/albumin molar ratios to yield three ranges of Bf concentrations (10–20, 40–50, and 80–90 nM). Cell viability was ascertained following a 2h exposure to bilirubin. This time interval was based on a time course curve made for each cell line.

We compared the susceptibility of four cell lines to bilirubin toxicity: HeLa cells (20) and mouse embryo fibroblasts (21), previously used for bilirubin studies in our laboratory, and two neuronal cell lines, SHSY5Y neuroblastoma cells and striatal precursor 2a1 cells.

Effect of sulfadimethoxine on Bf and cell viability.

Bf was measured under the optimal growth conditions (15% FCS) for SHSY5Y, with or without 250 μM sulfadimethoxine, a strong binding competitor for UCB (22). BT ranged from 3 to 32 μM with total BSA concentration of 54 μM. Similar Bf in the presence and absence of sulfadimethoxine were established by using much lower BT concentrations in the presence of sulfadimethoxine. Control systems contained 250 μM sulfadimethoxine and the same volume of DMSO, but no UCB. Cells were incubated for 2 h and MTT assay then performed.

Effect of light exposure on Bf and toxicity.

The effect of light (1000 lux) on Bf was evaluated using 15% FCS and 32 μM UCB in F12/EMEM medium. The solution flask was placed 10 cm below a GE polylux XL F36W/840 fluorescent lamp (General Electric Company, UK) for 60 min, and Bf measured after warming to 37°C. SHSY5Y cells were then incubated for 2 h in medium that had or had not been exposed to light and viability was assessed by MTT assay.

Statistical analysis.

Results are expressed as mean ± SD of three experiments for each experimental condition. Statistical analysis was performed by ANOVA with Tukey test, using GraphPad InStat 3 software (San Diego, CA). Differences among the conditions were considered significant at p < 0.05.

RESULTS

Effect of different binders on Bf levels.

With all 3 types of albumin, Bf increased with increasing BT (Fig. 1). At a given BT, 30 μM BSA and 10% (vol/vol) FCS yielded comparable Bf values. These were roughly double the Bf obtained using the same concentration of HSA when BT was lower than 15 μM, consistent with the weaker binding affinity of BSA (13).

Figure 1
figure 1

Relationship of Bf to BT with four different albumin preparations. As follows: (▴) FCS 10%, (□) BSA 30 μM, (♦) HSA 30 μM, (▪) HSA 60 μM Data represent the mean ± SD of three independent experiments in triplicate.

Full size image

Effect of Bf on cell viability in different cell lines.

Figure 2 shows the changes in cell viability when the 4 cell lines were exposed to a Bf of 10–20, 40–50 and 80–90 nM for 2 h. A constant Bf with varying BT was obtained by using albumin preparations with different binding characteristics, as shown in Fig. 1. Irrespective of BT, similar toxicity was observed at similar Bf obtained by varying the type of albumin and the UCB/albumin ratios, or displacing UCB with sulfadimethoxine (Fig. 3).

Figure 2
figure 2

Effect of Bf on viability of four different cell lines, assessed by MTT test. 2a1 (A), HeLa (B), MEF(C) were incubated with DMEMHG but SHSY5Y cells (D) were incubated in F12/EMEM medium culture for 2 h at various Bf as indicated. HSA 30 (♦) or 60 μM (▪), BSA 30 μM (□) and FCS 10% (▴) were the bilirubin binders. Data represent the mean ± SD of three separate experiments in triplicate. (*) p < 0.05.

Full size image
Figure 3
figure 3

Effect of sulfadimethoxine on Bf and cell viability. SHSY5Y cells were incubated in F12/EMEM with 15% vol/vol FCS (albumin concentration = 54 μM) with (black bars) or without (open bars) 250 μM sulfadimethoxine. BT was adjusted between 3 to 32 μM to yield similar Bf in the presence and absence of sulfadimethoxine, using much lower BT concentrations in the presence of sulfadimethoxine. Control systems contained 250 μM sulfadimethoxine but no UCB. Viability was assessed by the MTT assay after 2 h of incubation with UCB. Data represent the mean ± SD of three independent experiments performed in triplicate.

Full size image

Although the susceptibility to damage differed among the 4 cell lines, their viability after two hours of incubation with UCB generally decreased as Bf increased (Fig. 2). SHSY5Y cells were most sensitive to UCB with about a 20% loss of viability at Bf = 40–50 and 30% loss at Bf = 80–90 nM (p < 0.05 for each). HeLa cells were likewise significantly affected at these concentrations, albeit less severely. In contrast, 2a1 cells showed a modest decrease in viability only at a Bf = 80–90 nM (p < 0.05) while MEF cells showed a trend toward decreased viability that did not achieve statistical significance. Similar results were observed after 1 h exposure to UCB: viability of SHSY5Y cells was already significantly impaired at Bf = 40 nM, whereas HeLa, 2a1 and MEF cells showed unimpaired survival even at Bf = 80–90 nM (data not shown). Longer exposure (6 h) to a Bf = 80–90 nM resulted in a further increase in cytotoxicity in SHSY5Y cells (40 ± 3% decrease in viability) and the appearance of a cytotoxic effect in the other cell lines (20 ± 5 for HeLa, 15 ± 4% for 2a1 and 16 ± 3 for MEF).

Effect of sulfadimethoxine on Bf and on viability of SHSY5Y cells.

As shown in Fig. 3, Bf increased about 5 fold and 3 fold in the presence of sulfadimethoxine, when the bilirubin/FCS molar ratios were 0.11 and 0.22 respectively, confirming the ability of the drug to displace UCB from albumin. Sulfadimethoxine in the absence of UCB had no effect on cell viability (100 ± 9.1 control versus. 94.4 ± 12.4 treated cells, NS). Loss of cell viability was comparable at any given Bf level, irrespective of BT (low in the presence of sulfadimethoxine) or BT/albumin molar ratio.

Effect of Bf on cell viability without albumin.

We exposed SHSY5Y cells for 2 h to BT ranging from 10 to 1000 nM, without albumin. Thus, all UCB in the medium is initially “free” (Bf = BT). Cell viability remained unchanged up to Bf = 500 nM (Fig. 4). In contrast, cells incubated in the presence of 15% vol/vol FCS and increasing concentrations of UCB to achieve Bf varying from 10–20 to 140–160 nM showed cytotoxicity; at a Bf of 140–150 nM, viability was 40–50% lower (p < 0.001).

Figure 4
figure 4

Effect of UCB without albumin on cell viability. SHSY5Y cells were incubated with UCB in F12/EMEM medium without (Δ) or with 15% vol/vol FCS (▴), with BT adjusted to yield comparable Bf in both systems. Viability was assessed by the MTT assay after 2 h of incubation with UCB. Data represent the mean ± SD of three separate experiments performed in triplicate.

Full size image

Effect of light on Bf and cell viability. Addition of 32 ± 1.5 μM UCB to 15% FCS (BSA 54 μM) resulted in a Bf of 164 ± 16 nM which decreased to 136 ± 13 nM following 60 min exposure to 1000 lux (p < 0.01). BT decreased to 27.3 ± 0.9 μM (p < 0.01) and the calculated apparent binding constant, Kf, (13) assuming one binding site on BSA (23), decreased from 8.98 ± 0.8 × 106 to 7.44 × 106 ± 0.7 M−1, (p < 0.06). Viability of SHSY5Y cells incubated for 2 h decreased 45.9 ± 6.8% in nonirradiated media, but only 34.6 ± 6.9% in irradiated media (p < 0.05). The improved viability after irradiation was appropriate for the lower Bf found. In neither condition did Bf change significantly during incubation (Bf = 164 ± 16 versus. 155 ± 13 in non irradiated solution and 136 ± 14 versus. 130 ± 10 after irradiation). This was anticipated, since uptake of unbound UCB by the cells during incubation was buffered by release of Bf from the reservoir of albumin–bound UCB.

DISCUSSION

In this study we demonstrated variable sensitivity of four different cell lines to bilirubin toxicity. Loss of viability was dependent on the free bilirubin level (Bf) but not the total bilirubin concentration (BT), whether or not the Bf is varied by using preparations of albumin with different binding affinities, or by displacement of UCB from albumin by sulfadimethoxine. The results obtained with sulfadimethoxine confirm the observation that viability of cultured 8402 human cells correlated (r2 = 0.94) with the square root of Bf level, with or without sulfisoxazole (24). SHSY5Y cells, a neuronal model for studies of Alzheimer's disease (25), are more sensitive to UCB cyotoxicity than the 3 nonneuronal cell lines tested. The response of SHSY5Y cells to bilirubin is similar to other neuronal or astrocytic cell lines reported in the literature (11). Interestingly, this greater susceptibility of SHSY5Y cells to UCB toxicity is manifested both at lower Bf and at shorter incubation times compared with the other three cell lines tested. Although the molecular pathogenesis of bilirubin-induced neuronal cell injury is not completely understood, cell damage at any given Bf is affected by a variety of mechanisms (26) including export of UCB by ABC transporters, binding of UCB by cytosolic proteins, cellular oxidation and conjugation of UCB.

We observed a marked difference in dose-response when SHSY5Y cells were exposed to UCB in the absence of albumin. The Bf required for a comparable decrease in viability was more than double that observed with albumin present. Nelson reported similar findings with L929 cells (10). These results can be explained in part by self-aggregation of unbound UCB and by the reservoir function of the albumin-bound UCB. The true determinant of UCB cytotoxicity is the intracellular content of UCB. In the absence of the reservoir of albumin-bound UCB, the small amount of unbound UCB in the extracellular fluid in our experimental system (150 pmol at a Bf (= BT) of 150 nM) is quickly consumed, so that a sufficiently high toxic intracellular level of UCB is not reached. Unfortunately, due to extremely low concentration, it was impossible to measure the “equilibrium” extracellular BT in the absence of albumin, or to compare the intracellular UCB concentrations with and without albumin. With albumin present, UCB taken up from the extracellular pool of unbound UCB is instantaneously replaced by dissociation of bound UCB from the albumin, thus maintaining a constant Bf. In the presence of 15% FCS, a Bf of 150 nM requires a total bilirubin content of 20,000 pmol i.e., more than 130 times greater than that found at a comparable Bf in the absence of FCS. Binding to albumin also prevents the pigment from aggregating and precipitating and stabilizes the pigment against oxidation (27). Collectively these data indicate that in addition to the Bf, the total amount of UCB available for uptake is also important to predict cellular damage. In vivo, the presence of a large reservoir of albumin-bound UCB renders this consideration irrelevant.

Application of phototherapy to jaundiced infants produces photo-isomers and photo-oxidative derivatives of UCB. The water soluble photo-isomers are thought to be nontoxic, although confirming experimental evidence is lacking, and they cannot be discriminated from native UCB (bilirubin IXαz,z) by usual assays for BT. There is evidence that these isomers bind less avidly to albumin (28), and concern has been raised that they will be oxidized faster than bilirubin IXαz,z, and yield misleading values of Bf (29,30). Our data support previous observations that photo-products do not significantly alter the measurement of Bf (31,32), either by displacing native UCB from the binding site and/or by having a lower binding affinity to the albumin or the peroxidase enzyme. Regardless of mechanism, the predictable relationship between measured Bf and toxicity in medium exposed to intense light supports the validity of the peroxidase assay in evaluating risk for bilirubin toxicity when photo-isomers are present.

The presence of serum or albumin in culture media is often necessary for optimal cell growth. The use of BSA or FCS rather than HSA in studies of UCB cytotoxicity in vitro is supported by several reasons among which the most relevant are: 1) BSA is much less expensive than HSA; and 2) BSA has a lower affinity constant for UCB, requiring a much lower BT to achieve any given Bf level (13). Using the peroxidase method (12), Bf can be easily measured without dilution of culture medium in any study where BSA or FCS is used as a UCB reservoir (13). Measurement of Bf in all in vitro experiments would greatly facilitate interpretation and comparison of dose-response studies conducted in different laboratories under different conditions.

This study supports the “free bilirubin theory” which hypothesizes that bilirubin toxicity occurs when unbound bilirubin enters the brain and binds to intracellular structures (1). A recent outcome study of newborns exposed to marked hyperbilirubinemia (4) emphasizes the limitation of BT as a predictor of neurologic outcome. Verification of Bf as the critical determinant of toxicity in vitro reinforces the proposal to assess the risk for kernicterus by measuring Bf in hyperbilirubinemic newborn plasma (3), rather than depending solely on BT, as currently recommended by the American Academy of Pediatrics (2).