Delayed Response to Phenobarbital Treatment of a Crigler-Najjar Type II Patient with Partially Inactivating Missense Mutations in the Bilirubin UDP-Glucuronosyltransferase Gene

The potential neurotoxin, bilirubin, is produced in abundance daily from senescent red blood cells, and its detoxification is dependent solely on the bilirubin UDP-glucuronosyltransferase (transferase) enzyme. This lipophilic-behaving heme derivative is rendered water soluble and is excreted through conjugation to glucuronic acid by the transferase. Developmentally, bilirubin transferase is often delayed. As a consequence, neonates are often exposed to episodes of high serum bilirubin concentrations that may lead to deposits in the central nervous system and cause kernicterus (neurotoxicity) (1). This occurrence is a major concern for neonatologists and clinicians because of the possibility for permanent damage to the central nervous system of neonates. There are inheritable diseases or syndromes involving bilirubin transferase, which reflect three levels of severity. In Crigler-Najjar (CN) disease type I (2), there is essentially no detectable bilirubin transferase activity, severe serum unconjugated hyperbilirubinemia (<20 mg/dl), and a high risk for kernicterus. The less severe CN disease type II involves intermediate bilirubin transferase activity, an intermediate hyperbilirubinemia (<20 mg/dl), and susceptibility to neurotoxicity due to bilirubin accumulation (3). Although occurrence of CN-I and CN-II is rare, a substantial 5% of the population classified as having Gilbert's disease (4) are often asymptomatic but have mild hyperbilirubinemia (∼1.0-1.5 mg/dl) caused by a more modest reduction in bilirubin transferase activity (5). Normal bilirubin levels range between 0.5 and 0.75 mg/dl. Typically, patients with CN-II, unlike those with CN-I, respond to phenobarbital treatment (6) by manifesting a reduction in serum bilirubin levels. Similar to CN-II patients, some with Gilbert's disease have responded to phenobarbital treatment (7). Patients with CN-II are distinguished from those with CN-I by a response to phenobarbital treatment. We report here that the high hyperbilirubinemia in an until the child reached 7 months of age, when a infant with CN-II did not respond to phenobarbital treatment until the child reached 7 months of age, when a second treatment was undertaken. Because of this delayed response, the child was initially incorrectly diagnosed as having CN-I and was listed for liver transplant surgery. To determine the genotype of the patient, we examined the bilirubin transferase UGT1A1 gene located in the UGT1A complex locus (8). CASE REPORT A 2-week-old African-American infant (whom we will call TS) with CN was admitted to the Children's Hospital of Wisconsin for treatment of jaundice, irritability, and vomiting. He was the product of a full-term, uncomplicated pregnancy and delivery. He was healthy appearing but deeply jaundiced. The liver margin was palpable but not the spleen. Total bilirubin was 32.7 mg/dl, conjugated bilirubin was 0.2 mg/dl, hemoglobin was 14.7 mg/dl, and the Coombs test result was negative. Two hours after intravenous treatment with antibiotics, fluids, and phototherapy, bilirubin was 24.4 mg/dl. Five days later, bilirubin had stabilized at 12.9 mg/dl, and the child was discharged home. After 3 days, the infant was again admitted to the hospital with bilirubin of 18.6 mg/dl and treated again with phototherapy and intravenous fluids. Phenobarbital (2 mg/kg) was administered every 12 hours. A week later the bilirubin concentration was 11.8 mg/dl, and phototherapy was discontinued. The following day bilirubin was 13.5 mg/dl, and the diagnosis of Crigler-Najjar type I was made. Phenobarbital treatment was discontinued, and phototherapy was resumed. The child was listed for liver transplant; but as described later, this was a temporary provision. The bilirubin level remained stable between 14 and 15 mg/dl with 12 hours of phototherapy daily. When the patients was 7 months of age, 2 mg/kg phenobarbital was again administered every 12 hours; 1 week later the bilirubin concentration was 4.6 mg/dl. Phototherapy was discontinued, and the correct diagnosis of CN type II was made. The bilirubin level stabilized and has remained between 8 and 9 mg/dl. The child is now 3 years old and thriving. To establish the genetic defect in the patient, we analyzed genomic DNA prepared as already described (9) from leukocytes. The sources of reagents used to carry out recombinant DNA techniques has been published (9,10). Exons of the UGT1A1 gene encoding the bilirubin UDP-glucuronosyltransferase were amplified by polymerase chain reaction (PCR) and subcloned for sequencing as described (9,11). A coding region alteration exists on each allele of the UGT1A1 gene. An A to G transition changed Met to Val at codon 310 (M310V; Fig. 1A), and a transition of T to C at codon 431 replaced Ile with Thr (1431T; Fig. 1B). UGT1A1*34 and UGT1A*35 represent the M310V and 1431T mutations, respectively.FIG. 1: Comparison of nucleotide sequences in a wild-type UGT1A1 (HUG-Br1) cDNA with that of UGT1A1*34 (M310V) and UGT1A1*35 (I431T) mutations. Sequencing reactions representing the wild-type exon 2 (top left) and that from the patient with Crigler-Najjar disease type II are shown as UGT1A1 and UGT1A1*34, respectively. A normal exon 4 (bottom left) is compared with that from the patient coding for the UGT1A1*35 (bottom right) mutation. The solid circle represents the substituted nucleotide in the patient genome.A mutant cDNA representing each mutation was constructed in the pSVL-based UGT1A1 cDNA unit as previously described (10). Point mutations were introduced in the cDNA at amino acid positions 310 and 431, as detailed. The primer sets were as follows: M310V, sense 5′-TTTGGGATCAGTGGTCTCAGAAA, and antisense 5′-TTTCTGAGACCACTGATCCCAAA-3; and, I431T, sense 5′-AAAAGCAGTCACCAATGACAAA-3′, and antisense 5′-TTTGTCATTGGTGACTGCTTTT-3′. The outside primers for M310V were sense OP170, 5′-CAGGGCGGACGCCCACTTGT-3′, and antisense PXAS6, 5′-TAAACACCATGGGAACC-3′. In the case of I431T, the outside primers were sense P2S4, 5′-CTGTGCGACGTGGTTTA-3′ and antisense AG2, 5′-CTGTCTGCACGTCCTCTGAA-3′. The alleles specifying the M310V and I431T substitutions were designated UGT1A1*34 and UGT1A1*35, respectively, and the corresponding mutant proteins were designated UGT1A1*34 and UGT1A1*35. COS-1 cells were transfected with the pUGT1A1, pUGT1A1*34, or pUGT1A1*35 plasmid. Newly expressed wild-type and mutant transferase proteins were radiolabeled, immunoprecipitated, and analyzed by sodium dodecyl sulfate-polyacrylamide gel electrophoresis, as previously described (9,10). Transferase proteins are shown in Figure 2. Similar amounts of specific protein in the homogenates of the different transfected COS-1 cells were established (11) and used to determine the relative levels of bilirubin glucuronidation, as previously described (12,13).FIG. 2: Immunocomplexes of the UGT1A1 protein and its mutations, UGT1A1*34 and UGT1A1*35, synthesized in COS-1 cells. Cells were transfected with the pUGT1A1, pUGT1A1*34, or pUGT1A1*35 expression unit and incubated for 72 hours. The proteins were radiolabeled as described (11). Washed product was separated on a sodium dodecyl sulfate-polyacrylamide gel during electrophoresis and dried. Specific bands were quantitated by scanning on a phosphoroimager, as described (13).The total bilirubin-glucuronide generated by the UGT1A1*34 (M310V) and UGT1A1*35 (I431T) proteins is shown in Figure 3; activity remaining, compared with that of the wild type UGT1A1, was 26%/51% and 61%/83%, respectively, at pH 6.4/pH 7.6. Earlier, we showed that wild type activity is two to three times higher at pH 6.4 than at pH 7.6 (12). We do not know, however, the significance of the activity seen at the two pH values.FIG. 3: Glucuronidation at pH 6.4 and pH 7.6 catalyzed by the UGT1A1*34 and UGT1A1*35 mutant bilirubin transferase isozymes uncovered in the patient with Crigler-Najjar disease type II. COS-1 cells were transfected with either pUGT1A1, pUGT1A1*34, or pUGT1A1*35 expression unit and incubated for 72 hours. The proteins were radiolabeled as described (11). Washed product was separated on a sodium dodecyl sulfate-polyacrilamide gel during electrophoresis and dried. Specific bands were quantitated by scanning on a phosphoroimager, as described (13). The same amount of specific protein for UGT1A1, UGT1A1*34, and UGT1A1*35 was used to determine bilirubin glucuronidation at pH 6.4 and pH 7.6 as described in Methods (13).DISCUSSION We have identified two missense mutations in a CN-II patient. Each mutation when placed in the cDNA and expressed in the COS-1 cells exhibited reduced bilirubin glucuronidation at pH 6.4 and pH 7.6. On average, 39% to 72% normal activity remained with the mutant proteins. Earlier, we uncovered a 1294T missense mutation in another CN-II patient who also exhibited a delayed response to phenobarbital treatment. Although the retention of in vitro activity by the mutant proteins can account for the diagnosis of CN-II, it is not obvious why the induction of either or both of these proteins had a delayed response to phenobarbital treatment. In agreement with this finding, we report that the hyperbilirubinemia in another CN-II patient did not respond to phenobarbital treatment until the patient was 3 months of age (14). That patient is also compound heterozygous and has a 1294T missense mutation on one allele and a TA insertional mutation at the TATA box of the second allele of the bilirubin transferase gene (15). I294T (UGT1A1*33) exhibited 40% to 55% normal activity. Repeating the phenobarbital treatment to verify the diagnosis had no clinical justification (SLW). The rationale for a second treatment for TS was based on the success reported by Rubaltelli (14). In personal communications, Rubaltelli also indicated that there was no clinical basis for his second attempt to elicit a response to phenobarbital treatment. Therefore, it is important to emphasize the necessity for repeated treatment to prevent erroneous diagnoses, especially when the alternative treatment is often liver transplantation. Because the first child (14,15) is Italian (whom we will call SM) and lives in Italy compared with TS who is African-American and lives in Minnesota, it is likely that the genetic mutations, rather than environment factors, are the most influential in the response to phenobarbital treatment, although the mechanism is unknown. Rubaltelli indicated that he has seen five patients with CN-II; in three, hyperbilirubinemia showed a delayed response to phenobarbital treatment. TS was the single CN-II patient seen by SLW. Although there are other reports (16-19) concerning genetic defects in CN-II patients, we found no evidence of a delayed decrease in hyperbilirubinemia in response to the administration of phenobarbital. There was a report (20) of orally administered phenobarbital that revealed a better response at an older age, which could have been related, however, to a difference in the absorption of phenobarbital. Thus, it may be necessary to develop a uniform treatment regimen for the diagnosis of the CN-II disease to assure the appropriate use of liver transplant therapy. The genotype of patients exhibiting a delayed response to the barbiturate is worth documenting. Because phenobarbital causes increased bilirubin transferase activity through increasing mRNA and protein levels (21,22), it is possible that the basis of the delay is a simple developmental phenomenon peculiar to the patient, to the particular mutation in the bilirubin transferase protein, or to both. The delay could reflect slow-to-develop secondary structural changes necessary for activity. A study concerning a rat UDP-glucuronosyl-transferase showed convincingly that at least one transferase form undergoes homodimerization and that mutations in the amino terminus of this isozyme affected the dimerization (23). Of the three mutations seen in the CN-II patients described or referenced here, M310V and I294T are in the amino terminus, and I431T is in the carboxyl terminus. Whether homodimerization occurs for the bilirubin transferase protein molecule has not been established. It is possible that a high cellular concentration of the compromised monomeric protein is required for dimerization before a stable and partially active homodimeric protein is present in sufficient levels to have a measurable affect on serum bilirubin levels. The observations detailed in this report clearly require further investigation and point out the necessity for repetitive treatment of patients who do not respond to phenobarbital treatment, as was the case for TS in this study and for SM (14,15). Both required multiple phenobarbital treatments for several months before the correct diagnosis of CN-II was made. Because the hyperbilirubinemia seen in most CN-II patients responds to phenobarbital treatment without delay, it remains to be seen whether other patients with the mutations, other than the two or three patients discussed here, have a delayed response to phenobarbital. It is possible that slowed secondary changes in the mutant bilirubin transferase protein structure could account for the delay in the manifestation of enzyme activity.