Despite a wealth of studies on hyperbilirubinemia and brain injury, many fundamental questions regarding the mechanisms of bilirubin-induced cell damage and cell death remain unanswered, their details incompletely understood, and their potential clinical relevance not fully characterized (1–5). Does bilirubin neurotoxicity reflect interference with single or mul-tiple pathways (2,3)? Is there a unifying mechanism to explain bilirubin intoxication (1,2)? Why are only selected cell types (and not all neurons) readily “poisoned” by bilirubin (5)? What is the sequence and relative importance of toxic events in bilirubin induced cell injury and cell death (5)?

Neuronal cell damage in kernicterus is classically described as eosinophilic necrosis, but this literature largely predates recognition of programmed cell death or apoptosis and on closer inspection describes a spectrum of histologic and cytologic alterations across time of which necrosis may be but a final common pathway or end point (6). Of particular interest in kernicteric human neonates are early, subacute neuropathologic changes akin to apoptosis including nuclei with increased nuclear density and a lack of inflammation (6), the latter a hallmark of apoptosis. Thus, apoptosis, an active energy (ATP) requiring, highly organized, controlled process of cell self-destruction may represent an important means of bilirubin-induced cell death particularly during the early phases of injury. In contrast, later neuropathologic changes of kernicterus are characterized by a predominance of neuronal cell loss, prominent reactive astrocytosis, and necrosis (6).

Hanko and colleagues in a compelling paper in this issue of Pediatric Research determined the modes of cell death in human NT2-N neurons and their temporal evolution following exposure to various concentrations of unconjugated bilirubin (UCB) in vitro (7). As the investigators hypothesized and others as cited by the authors suggest, apoptosis and necrosis may represent ends of a cell death spectrum where the magnitude of the insult determines whether a cell undergoes apoptosis or necrosis (7–9). Indeed, Hanko and coworkers demonstrated that increasing UCB concentration from <1 to 250 μM at an UCB to BSA ratio of 1.5 caused a dose dependent increase in NT2-N neuronal cell death in vitro. Moreover they observed that: 1) low concentration [5 μM] UCB exposures were associated with apoptosis as evidenced by fragmented nuclei, DNA laddering, and cleavage of the DNA repair enzyme and death substrate poly (ADP) ribose polymerase (PARP); whereas 2) moderate [10 μM] to high [≥ 25 μM], UCB concentrations were associated with morphologic nuclear variants typical of necrosis and an absence of DNA laddering and PARP cleavage. Consistent with these findings are the observations of Silva and colleagues who noted that although short duration (4 h) UCB exposures resulted in similar degrees of apoptosis and necrosis in rat astrocytes in vitro, necrosis predominated over apoptosis at higher UCB concentrations and longer exposure times (10). These reports are compatible with the notion that the magnitude and duration of insult, in this case UCB concentration, is critical to the death pathway activated and that stimuli that lead to apoptosis are typically milder than those that cause necrosis.

Others also suggest that apoptosis plays a role in bilirubin-induced cytotoxicity in vitro (10–15) in murine neurons, astrocytes, and capillary endothelial cells, and that programmed cell death in this context appears to be caspase dependent (although caspase independent alternative cell death mechanisms cannot be entirely excluded at this time). Such evidence includes activation of caspase 3 (15), a key distal effector of apoptosis; PARP cleavage (11,15); decreased mitochondrial transmembrane potential (15); increased mitochondrial cytochrome c release (15); DNA fragmentation (11); terminal deoxynucleotididyltransferase-mediated dUTP-biotin nick end labeling (TUNEL) (10); Hoescht nuclear staining (15), and transmission electron microscopic ultrastructural nuclear changes (11). This growing body of literature suggests that enhancing our understanding of bilirubin-induced apoptosis is an important goal to more fully characterize the pathogenesis of bilirubin neurotoxicity. Such insights could hold potential therapeutic benefit given that the cascade of events in programmed cell death can be interrupted.

The role of apoptosis in bilirubin-induced cell injury is of increasing interest as well in light of recent findings suggesting that ATP-binding cassette (ABC) transport proteins may limit the degree of bilirubin-induced apoptosis in vitro (16,17). Gennuso and coworkers report that inhibition of murine astrocyte multidrug resistance associated protein-1 (MRP-1), an efflux pump of glutathione, xenobiotics, and UCB (18), during UCB exposure in vitro, is associated with increased levels of astroglial apoptosis (16). A preliminary report suggests an analogous cellular pro-apoptotic effect of MDR1 P-glycoprotein, an ABC transporter of xenobiotics and possibly UCB (19), inhibition in vitro (17). In this regard, both MRP-1 and MDR1 P-glycoprotein are expressed in the developing human CNS, MDR1 P-glycoprotein in brain capillary endothelial cells and perivascular astrocytes of the blood-brain barrier, MRP-1 and MDR1 P-glycoprotein in choroid plexus epithelium of the blood-CSF barrier, and both in cells of the CNS parenchyma not associated with blood vessels including astrocytes within white matter, large pyramidal cells of the brainstem and thalamus, and Purkinje cells of the cerebellum (20). These findings taken together suggest that ABC transport proteins may play a neuroprotective role against bilirubin-induced apoptosis. Whether these effects are mediated by directly limiting the cellular uptake of UCB, extruding UCB before it has a chance to exert its cytotoxic effect, or by a mechanism independent of its bilirubin transport function (21) is unclear but merits further study. Similarly, whether up-regulation of ABC transporter expression in cells of the CNS may further attenuate bilirubin-induced injury and cell death is worth exploring.

It will be important, however, to substantiate that bilirubin-induced apoptosis observed in vitro is a phenomenon that occurs in vivo. Programmed cell death depends on the confluence of a number of factors and may be different in cell lines exposed to bilirubin in vitro compared with cells in vivo. To date reported evidence for bilirubin induced apoptosis in vivo is limited to morphologic findings under light microscopy in the Gunn rat animal model of neonatal jaundice and kernicterus, more specifically, apoptotic nuclear profiles in cerebellar Purkinje cells in homozygous hyperbilirubinemic animals (22). These animals were treated with sulfadimethoxine (200 mg/kg ip) to enhance the blood concentration of free UCB, increase brain tissue bilirubin accumulation, and thereby neuronal cell injury (22). Although nuclear morphologic features are considered reliable criteria for defining apoptosis, complementary studies using biochemical or molecular assays of apoptosis have yet to be reported in the Gunn rat model of kernicterus or in human kernicteric postmortem tissue samples. Specific markers for apoptosis including 1) caspase activation, e.g., activated caspase-3 using either immunohistochemical or Western immunoblotting techniques, and 2) detection of apoptotic (annexin-V positive; propidium iodide negative) and necrotic (propidium iodide positive) cells using flow cytometry would help to substantiate that bilirubin induced apoptosis is a relevant phenomenon in vivo. An important challenge in efforts to document bilirubin induced apoptosis in vivo will be the rapid clearance of apoptotic cells from tissue, i.e. only a few cells are likely to be observed undergoing apoptosis at a single time point. Therefore, assessment at multiple time points may be necessary to detect programmed cell death in vivo. Even with such an approach the rapid phagocytosis of apoptotic cells in vivo will make an accurate assessment of the contribution of apoptosis to overall cell death in vivo difficult to quantify (may in fact preclude it). Documentation of bilirubin induced apoptosis in vivo, however, is necessary to substantiate the potential clinical relevance of apoptosis to kernicterus even if the results of such studies will likely underestimate its contribution.