Complete Rescue of Lethal Albino c 14CoS Mice by Null Mutation of 4-Hydroxyphenylpyruvate Dioxygenase and Induction of Apoptosis of Hepatocytes in These Mice by in VivoRetrieval of the Tyrosine Catabolic Pathway*

Hereditary tyrosinemia 1 (HT1) is characterized by progressive liver damage, from infancy, and by a high risk for hepatocellular carcinoma. HT1 is due to mutations in the fumarylacetoacetate hydrolase gene Fah, encoding the last enzyme in the tyrosine catabolic pathway. Lethal albino deletionc 14CoS mice and mice with target-disruptedFah are models for HT1, but they die in the perinatal period, albeit with a different phenotype from that seen in HT1 in humans. We first asked whether homozygous null mutation of the 4-hydroxyphenylpyruvate dioxygenase gene Hpd could rescue the homozygous c 14CoS mice (c 14CoS /c 14CoS orFah −/−). The double mutantFah −/− Hpd −/− mice appeared normal, at least until age 18 months, and there was no evidence of liver disease, findings that facilitated examination of the effect of Fah −/− on mature and unmodified hepatocytes in vivo. The hepatocytes ofFah −/− undergo rapid apoptosis, and acute death follows. Essentially the same phenomena were observed whenFah −/− Hpd −/− mice were administered homogentisate intraperitoneally. These changes in liver pathology in Fah −/− Hpd −/− mice after the administration of homogentisate were associated with massive urinary excretion of succinylacetone. These results suggest that accumulation of fumarylacetoacetate, maleylacetoacetate, or succinylacetone seems to trigger the endogenous process of apoptosis in hepatocytes that lack fumarylacetoacetate hydrolase activity. This apoptosis may be related to the development of hepatocellular carcinomas seen in HT1 patients and pharmaceutically treated fumarylacetoacetate hydrolase-deficient mice.

Hereditary tyrosinemia 1 (HT1) 1 is characterized by progres-sive liver damage, from infancy, and by a high risk for hepatocellular carcinoma (1,2). The cause of HT1 is mutations in the fumarylacetoacetate hydrolase gene Fah, encoding the last enzyme in the tyrosine catabolic pathway (1,3). In the acute form of the disease, the liver is sometimes nonfunctional by the time the infant is 2-3 months old (1,2). In contrast, in the chronic form, the progression of liver damage is slower, and liver function can continue for several years (1,2). Although the recent prescription of NTBC, an inhibitor of 4-hydroxyphenylpyruvate dioxygenase encoded in the gene Hpd, has been reported to be effective in delaying progression of the liver disease (4), these patients require liver transplantation (2).
Pathological features of the liver in HT1 patients are characteristic, but not diagnostic (1,2). In the chronic form, established liver cirrhosis and often hepatocellular carcinomas are present. In the acute form of the disease, the hepatocytes are pleiomorphic with fatty metamorphosis and cholestasis, and there is pseudogland transformation. These features are indistinguishable from those seen in cases of galactosemia and hereditary fructose intolerance, thereby implying a common mechanism. While such features are presumably related to events leading to hepatocyte injury and death, the actual process of liver damage has not been defined in any cases.
Lethal albino deletion c 14CoS mice and mice with targetdisrupted Fah are models for HT1 (5)(6)(7)(8)(9)(10)(11)(12), but these mice die in the perinatal period, bearing a phenotype different from that seen in HT1 patients (5)(6)(7)(8)(9). Studies on these mice revealed impairment of expression of developmentally regulated genes that are essential for liver functions (6 -11). These abnormalities led to neonatal death probably due to hypoglycemia (8). The hepatocyte injuries seen in HT1 patients have not been evidenced in these mice, but ultrastructural investigations on these lethal albino mice revealed altered membranous components of hepatocytes in the perinatal period (9). Heretofore, available mouse models of HT1 did not demonstrate progressive liver damage and did not allow for examinations of effects of the tyrosine catabolic pathway on mature and unmodified hepatocytes with the FAH defect.
In this study, we asked if homozygous null mutation of the Hpd gene would rescue the homozygous c 14CoS mice (c14CoS/ c 14CoS or Fah Ϫ/Ϫ ) (5,8,9,12). HPD catalyzes a complex reaction to form homogentisate from 4-hydroxyphenylpyruvate. The homozygous mutant Hpd allele from mouse strain III is expected to completely block tyrosine catabolism (13)(14)(15). As the formation of homogentisate exclusively depends on HPD activity, a complete block of tyrosine catabolism at this step would result in deprivation of homogentisate and its oxidative products. The outcome of the double mutants (Fah Ϫ/Ϫ Hpd Ϫ/Ϫ ) itself is important for evaluating treatment strategies (4) and for understanding the mechanism of the related carcinogenesis (1,16). We report here that blocking of tyrosine catabolism concealed the liver damage in lethal albino deletion c 14CoS mice. Retrieval of the tyrosine catabolic pathway in the double mutants by in vivo gene transfer or by injection of homogentisate resulted in apoptotic death of the hepatocytes. These observations provide new insights into the disease process of HT1. This model will aid in assessing and developing treatments for HT1 patients.

EXPERIMENTAL PROCEDURES
Generation of Double Mutants-Heterozygous c 14CoS (Fah ϩ/Ϫ ) mice (kindly provided by G. Schü tz) were crossed with homozygous III mice (Hpd Ϫ/Ϫ ), followed by the breeding of F1 ϫ homozygous III mice. Mice with the genotype Fah ϩ/Ϫ Hpd Ϫ/Ϫ were identified and used for the next breeding for generation of Fah Ϫ/Ϫ Hpd Ϫ/Ϫ mice. Heterozygotes for the Fah Ϫ allele (Fah ϩ/Ϫ ) were identified by Southern blots using the RN.Fd probe (a gift from G. Schü tz) as described (8,9). Homozygotes for the Fah Ϫ allele (Fah Ϫ/Ϫ ) were identified by the absence of exon 2 sequences and the presence of exon 8 sequences of the Fah gene after PCR amplification of regions containing each exon, respectively. The sequences of sense and antisense primers for PCR amplification of the regions of the mouse Fah gene were derived from the published cDNA sequence (12). Primer sets used for PCR amplifications were as follows (sense/antisense): a set for Fah gene exon 2, 5Ј-CCAAAGCCACGGAT-TGGTGT-3Ј/5Ј-TCATCGAAGACATGTTGATG; and a set for Fah gene exon 8, 5Ј-CACGAGACATCCAGCAATGG-3Ј/5Ј-CTGCTTTGGGTTTG-GCACCA-3Ј. The Hpd Ϫ allele was identified as described previously (15). DNA fragments were analyzed by electrophoresis on agarose gels and stained with ethidium bromide.
Immunoblotting of FAH and HPD-Liver samples were quick-frozen in dry ice and kept at Ϫ70°C until use. The tissues were homogenized in ice-cold 50 mM Tris-HCl buffer, pH 7.4, and centrifuged at 11,000 ϫ g for 15 min at 4°C. The supernatant (7.5 g of protein) was subjected to SDS-10% polyacrylamide gel electrophoresis. Immunoblotting of HPD and FAH was carried out with antiserum directed against recombinant human HPD and FAH, respectively. Rabbits were immunized with glutatione S-transferase-human FAH and glutatione S-transferase-human HPD for production of antibodies, respectively. 2 Preparation of human FAH cDNA (17) and HPD cDNA (18) was as described.
Other conditions for SDS-polyacrylamide gel electrophoresis and immunoblotting were as described (18). Amounts of protein were determined by the Bio-Rad dye binding method. Hepatic HPD activity was measured as described previously (19).
Recombinant Adenovirus-A replication-defective recombinant adenovirus, human adenovirus type 5 lacking the E1A, E1B, and E3 regions and bearing a human HPD expression unit in the E1-deleted region, was prepared as described (20). Transcriptional orientation of the expression cassette is leftward, opposite the original orientation of E1A and E1B genes. A BamHI-SalI fragment of human HPD cDNA was prepared as described (18) and inserted. Expression of HPD is driven by the CAG promoter, which consists of the cytomegalovirus immediateearly enhancer and the chicken ␤-actin/rabbit ␤-globin hybrid promoter containing rabbit ␤-globin 3Ј-flanking sequences, including a polyadenylation signal (21). The adenovirus was purified and titrated as described (22,23) and injected into the tail veins of the mice. Liver specimens were isolated before and after injection of the recombinant adenovirus, snap-frozen and kept at Ϫ80°C until use. A recombinant adenovirus, AdexCAGhOTC, which efficiently expresses human ornithine transcarbamylase in mouse liver when injected into the tail vein, was prepared as described previously (24). The effects of the intravenous administration of AdexCAGhOTC on the livers of C57BL/6 mice and ornithine transcarbamylase-deficient mice (spf ash ) were characterized (24).
Histological Examinations-For hematoxylin-eosin staining, livers were fixed in vivo by circulating 2% neutralized paraformaldehyde solution without treatment or after the administration of AdexCAGh-HPD or homogentisate. For electron microscopic investigation, a small piece of the liver was removed (12 h after homogentisate had been injected) and fixed in glutaraldehyde. The 3Ј-OH DNA ends generated by DNA fragmentation were detected by in situ terminal deoxyribonucleotidyltransferase-mediated dUTP-dioxygenin nick end labeling assay (25) using a kit from Oncor (Gaithersburg, MD) according to the manufacturer.
DNA Analysis-The livers of treated animals were excised at 24 h after the administration of the recombinant adenovirus (5 ϫ 10 8 pfu) or 12 h after injection of homogentisate and snap-frozen in liquid nitrogen. Nuclear DNA was prepared as described (25) and analyzed by electrophoresis on a 1.0% agarose gel.
Biochemical Tests of Blood and Urine-Serum levels of aspartate aminotransferase, alanine aminotransferase, alkaline phosphatase, bilirubin, blood urea nitrogen, and creatinine were measured using an automated analyzer. Blood tyrosine levels were assayed by high performance liquid chromatography using a column for neutral amino acids and dried blood spots as samples, as described (19). Urinary concentrations of succinylacetone were determined by stable isotope dilution gas chromatography-mass spectrometry assay as described (26).

Phenotype of the Double Mutants (Fah
Hpd allele (Hpd Ϫ ) is from mouse strain III (Hpd Ϫ/Ϫ ) (13-15), a model for human hereditary tyrosinemia type 3 (27) that is characterized by elevation of blood tyrosine and the absence of visceral injuries (13,27). There is a C to T transition at nucleotide ϩ7 on exon 7 of the Hpd gene on chromosome 5 of the III mouse, the result being premature termination of translation (15). This stop codon mutation was associated with the skipping of exon 7 in its mRNA (15). The transcript with the termination codon has not been obtained by reverse transcription-amplification of RNA from the liver (15). Thus, the amount of the 14-kDa polypeptide that was synthesized from mRNA with the stop codon mutation appears to be negligible. On the other hand, the abnormally spliced transcript can be translated into a 36-kDa protein lacking the central part of the subunit of HPD. This 36-kDa protein seemed to be unstable. As a result, protein immunologically related to HPD was absent in III mouse liver.
Genotyping of the pups revealed that the double mutants (Fah Ϫ/Ϫ Hpd Ϫ/Ϫ ) survived and grew well (Fig. 1, A and B). Analysis of liver biopsy samples confirmed that FAH and HPD proteins were absent in these mice (Fig. 1C). The clinical phenotype of Fah Ϫ/Ϫ Hpd Ϫ/Ϫ mice was indistinguishable from those of Fah ϩ/Ϫ Hpd Ϫ/Ϫ and Fah ϩ/ϩ Hpd Ϫ/Ϫ mice. The findings in liver sections from Fah Ϫ/Ϫ Hpd Ϫ/Ϫ mice ( Fig. 2A) were normal and similar to those from Fah ϩ/Ϫ Hpd Ϫ/Ϫ and III mice. Long-term investigations of Fah Ϫ/Ϫ Hpd Ϫ/Ϫ mice (12-18 months) revealed no evidence of hepatocellular carcinomas or preneoplastic lesions. Thus, the homozygous mutant Hpd allele from the III mouse not only rescued the lethal phenotype of Fah Ϫ/Ϫ mice, but also concealed the critical visceral phenotype of HT1.
Expression of HPD and Apoptosis of Hepatocytes in the Double Mutants-Phenotypically normal mice with inactive Fah provide a model in which the onset of visceral injuries of HT1 is controlled. At first, we attempted to retrieve the tyrosine catabolic pathway in these mice, as achieved by adenovirus-mediated in vivo expression of HPD in hepatocytes. Injection of the recombinant virus AdexCAGhHPD into the tail veins of the control mice (Fah ϩ/Ϫ Hpd Ϫ/Ϫ ) resulted in expression of HPD in the liver (Fig. 1D) and was accompanied by a reduction in blood tyrosine levels. No animal injected with 5 ϫ 10 8 pfu of Adex-CAGhHPD was lost in the control experiments using C57BL/6, Fah ϩ/Ϫ Hpd Ϫ/Ϫ , or III mice, except for one III mouse, which died within 1 h after the injection (Table I). The death of the III mouse seemed to be caused by direct reaction to the infusion. When 5 ϫ 10 8 pfu of AdexCAGhHPD was injected into Fah Ϫ/Ϫ Hpd Ϫ/Ϫ mice via the tail veins, blood tyrosine levels were reduced (from 21.4 Ϯ 3.4 to 3.2 Ϯ 1.6 mg/dl at 9 h after injection), and HPD protein appeared in the liver (Fig. 1D), but all the mice died within 30 h (range of 14 -30 h) (Table I). Although a recombinant virus carrying mutant HPD would be the best control virus, we used a recombinant adenovirus, AdexCAGhOTC, which expresses an unrelated protein. In control experiments, when AdexCAGhOTC (5 ϫ 10 8 pfu), which efficiently expresses human ornithine transcarbamylase in mouse liver when injected into the tail vein (24), was injected into the tail veins of Fah Ϫ/Ϫ Hpd Ϫ/Ϫ , III, or C57BL/6 mice, there were no significant changes in blood tyrosine levels.
Changes in the blood tyrosine levels in III and Fah Ϫ/Ϫ Hpd Ϫ/Ϫ mice were accompanied by an increase in hepatic HPD activities (Table II). The HPD activity was not detected in the original III mice (Fah ϩ/ϩ Hpd Ϫ/Ϫ ) and in Fah Ϫ/Ϫ Hpd Ϫ/Ϫ mice. The administration of AdexCAGhHPD into III mice resulted in a rapid increase in hepatic HPD activity, reaching levels in the wild type as early as 6 -12 h after administration (19). In the present study, livers were excised 6 h after the administration of AdexCAGhHPD, and HPD activities in the liver were measured. These results indicated that the activities in treated Fah Ϫ/Ϫ Hpd Ϫ/Ϫ mice were similar to those in treated Fah ϩ/ϩ Hpd Ϫ/Ϫ mice.
Fah Ϫ/Ϫ Hpd Ϫ/Ϫ mice treated with AdexCAGhHPD were definitely ill from 12 h after the injection, and mobility was reduced and appetite lost. These animals died within 30 h after injection of the recombinant virus (Table I). In control experiments, we injected AdexCAGhOTC into the tail veins of Fah ϩ/Ϫ Hpd Ϫ/Ϫ , III, or C57BL/6 mice; however, there were no significant changes in the clinical symptoms of these mice, and none were lost (Table I).
In liver sections obtained from the AdexCAGhHPD-treated Fah Ϫ/Ϫ Hpd Ϫ/Ϫ mice, massive numbers of hepatocytes were nonviable. There was no infiltration of inflammatory cells, but there were small areas of bleeding (Fig. 2b). The damaged cells showed evidence of chromatic condensation. When the liver sections were investigated for 3Ј-OH DNA ends generated by DNA fragmentation (25), ϳ15-25% of the hepatocytes in the observed regions were positive for the signals (Fig. 2c). Examination of nuclear DNA of the livers from the recombinant virus-treated Fah Ϫ/Ϫ Hpd Ϫ/Ϫ mice revealed fragmentation, with sizes corresponding to typical nucleosome units (Fig. 3). Thus, retrieval of HPD function in Fah Ϫ/Ϫ Hpd Ϫ/Ϫ mice resulted in apoptosis of hepatocytes. Treatment with AdexCAG-hOTC led to no significant changes in liver pathology in Apoptosis of Hepatocytes by Homogentisate in the Double Mutants-We next investigated the in vivo effect of an intermediate metabolite, homogentisate, on hepatocytes. Fah ϩ/ϩ Hpd Ϫ/Ϫ and Fah ϩ/Ϫ Hpd Ϫ/Ϫ mice showed no clinical symptoms or histological changes in the liver sections after injection of 10 -100 mg of neutralized homogentisate intraperitoneally (Table I). When 10 mg of neutralized homogentisate was injected intraperitoneally into 8-week-old Fah Ϫ/Ϫ Hpd Ϫ/Ϫ mice, all of the mice died within 16 h. (The mean dose given was 500 mg/kg. This amount of homogentisate corresponds to ϳ20 -25% of the daily intake of precursor amino acids, Phe ϩ Tyr.) Liver sections from homogentisate-treated Fah Ϫ/Ϫ Hpd Ϫ/Ϫ mice had massive numbers of dead hepatocytes with fragmentation of nuclei, as seen in the ultrastructural analysis (Fig. 2e). In addition, abnormalities in mitochondria were prominent. Approximately 20 -30% of the hepatocytes gave positive signals by in situ detection of DNA fragmentation (data not shown). The double mutant mice treated with 5 mg of homogentisate did not die (Table I); however, ϳ2% of the hepatocytes were positive for in situ detection of fragmentation of DNA (data not shown). Nuclear DNA from the livers of homogentisate-treated Fah Ϫ/Ϫ Hpd Ϫ/Ϫ mice showed typical fragmentation (Fig. 3). These apoptotic changes were nil in Fah Ϫ/Ϫ Hpd Ϫ/Ϫ mice treated with   AdexCAGhOTC. Liver sections from the homogentisate (10 mg)-treated animals with the genotypes Fah ϩ/Ϫ Hpd Ϫ/Ϫ , Fah ϩ/ϩ Hpd Ϫ/Ϫ , and Fah ϩ/ϩ Hpd ϩ/ϩ showed no significant changes. Blood Chemistry and Urinary Succinylacetone-To evaluate liver function in Fah Ϫ/Ϫ Hpd Ϫ/Ϫ mice before and after the administration of homogentisate or the recombinant adenovirus AdexCAGhHPD, serum levels of aspartate aminotransferase, alanine aminotransferase, alkaline phosphatase, bilirubin, and blood urea nitrogen were measured before and 12 h after administration. In Fah Ϫ/Ϫ Hpd Ϫ/Ϫ mice on a normal diet, serum levels of aspartate aminotransferase and alanine aminotransferase activities were essentially the same as in III mice (Fig. 4, A and B), suggesting that the liver and kidney functions of the double mutants on the normal diet were normal. After the administration of homogentisate, the levels of transaminases increased. The increases in aspartate aminotransferase were more prominent than those of alanine aminotransferase in the double mutants after the administration of homogentisate (Fig. 4, A and B). The serum levels of alkaline phosphatase in the untreated double mutants were similar to those in the controls and were relatively stable after the administration of homogentisate (data not shown). Increased levels of total bilirubin were observed in some Fah Ϫ/Ϫ Hpd Ϫ/Ϫ mice treated with homogentisate (Fig. 4C). The levels of blood urea nitrogen in some mice increased (Fig. 4D), suggesting that the renal functions of Fah Ϫ/Ϫ Hpd Ϫ/Ϫ mice treated with homogentisate were impaired. Essentially a similar, disordered liver function was seen when Fah Ϫ/Ϫ Hpd Ϫ/Ϫ mice were given 5 ϫ 10 8 pfu of the recombinant adenovirus AdexCAGhHPD, but the control recombinant adenovirus, AdexCAGhOTC, did not cause elevation of aspartate aminotransferase and alanine aminotransferase serum levels in the double mutants (data not shown).
A small amount of succinylacetone was detected in the pooled urine of untreated Fah Ϫ/Ϫ Hpd Ϫ/Ϫ mice (Table III). The concentration of succinylacetone was markedly increased in urine from Fah Ϫ/Ϫ Hpd Ϫ/Ϫ mice given homogentisate, whereas it remained at a low level when homogentisate was administered to III mice (Fah ϩ/ϩ Hpd Ϫ/Ϫ ). Blood chemistry tests of III mice treated with homogentisate indicated no significant changes in aspartate aminotransferase and alanine aminotransferase, thereby suggesting that the slightly increased formation of succinylacetone in the liver of III mice after administration of homogentisate did not lead to hepatocyte injury. This observation was consistent with findings that liver sections obtained from homogentisate-treated III mice indicated no pathological changes. The urinary excretion of succinylacetone in untreated Fah Ϫ/Ϫ Hpd Ϫ/Ϫ mice was slightly higher than in III mice, suggesting either that small amounts  .   TABLE III  Urinary excretion of succinylacetone from III and double mutant mice given or not given homogentisate Mice (6 -8 weeks old) were administered 10 mg of homogentisate or saline solution intraperitoneally, and urine samples were collected 6 h after the infusion. Pooled urine from mice (n ϭ 4) was analyzed for succinylacetone by stable isotope dilution gas chromatography-mass spectrometry assay. Mice  of homogentisate were formed in the livers of Fah Ϫ/Ϫ Hpd Ϫ/Ϫ mice or that homogentisate was included in the standard chow. The slight increase in succinylacetone seen in Fah Ϫ/Ϫ Hpd Ϫ/Ϫ mice did not correlate with increased serum levels of transaminases as none of the Fah Ϫ/Ϫ Hpd Ϫ/Ϫ mice fed the normal diet had significant changes in pathology, including carcinogenesis or premalignant lesions in the liver.

DISCUSSION
The c 14CoS mice that carried a homozygous deletion on chromosome 7 were normalized when the tyrosine catabolism pathway was completely blocked at the step of oxidation of 4-hydroxyphenylpyruvate by mutant HPD. This suggested that the impairment of expression of hepatocyte-specific and developmentally regulated genes seen in c 14CoS mice was due to oxidative product(s) of the homogentisate down the pathway of tyrosine catabolism or their derivatives. Because accumulation of homogentisate caused no abnormalities in the livers and kidneys of patients with alkaptonuria (28), who lacked homogentisate oxidase activity, oxidative product(s) of homogentisic acid are the primary candidates causing various abnormalities seen in c 14CoS mice. This is consistent with observations on c 14CoS mice that express FAH following transgenic manipulations: the c 14CoS mouse was rescued by transgenic expression of FAH (10). The FAH-deficient mice treated with NTBC, an inhibitor of HPD, survived but developed hepatocellular carcinomas (16). This may be due to possible differences in the extent of inhibition at the step of oxidation of 4-hydroxyphenylpyruvate. Our data suggest that liver carcinomas in the target-disrupted FAH-deficient mice could be caused by small amounts of the oxidative product(s) of homogentisate or their derivatives produced by an incomplete block of the oxidation of 4-hydroxyphenylpyruvate by NTBC.
The administration of homogentisate to Fah ϩ/Ϫ Hpd Ϫ/Ϫ , Fah ϩ/ϩ Hpd Ϫ/Ϫ , and Fah ϩ/ϩ Hpd ϩ/ϩ mice did not lead to changes in liver pathology or in liver function tests. Severe liver damage and death of animals after the administration of homogentisate were seen only in mice with the genotype Fah Ϫ/Ϫ Hpd Ϫ/Ϫ . Similarly, the administration of recombinant adenovirus that expresses human HPD in the liver led to death of the mice with elevation of serum transaminases and apoptotic death of hepatocytes. Hepatic failure is highly suspected as the cause of death of these treated double mutant mice.
An extremely high concentration of succinylacetone, which seemed to be derived from fumarylacetoacetate and maleylacetoacetate, was found in urine from Fah Ϫ/Ϫ Hpd Ϫ/Ϫ mice after the administration of homogentisate, but not in urine from homogentisate-treated Fah ϩ/ϩ Hpd Ϫ/Ϫ mice. Taken together, these results suggest that metabolites derived from homogentisate, i.e. fumarylacetoacetate (and maleylacetoacetate) or its derivatives, cause acute apoptotic death of mature and unmodified hepatocytes in Fah Ϫ/Ϫ Hpd Ϫ/Ϫ mice. Fumarylacetoacetate is strongly electrophilic (29); however, other substances have not been excluded as chemicals responsible for liver injury, including maleylacetoacetate, succinylacetoacetate, and succinylacetone. Accordingly, the development of hepatocellular carcinomas seen in pharmaceutically treated FAH-deficient mice (16) may be triggered by events related to fumarylacetoacetate and maleylacetoacetate and their derivatives. Although our investigation suggests that metabolite(s) derived from homogentisate are required for liver damage, further investigations will be needed to identify the substance responsible for the liver injury and hepatocellular carcinomas in FAH-deficient mice and HT1 patients.
The morphological abnormalities seen in hepatocytes of lethal albino deletion mice in the perinatal period (8,9) are likely to represent the very early stage of cellular damages. In vivo inhibition of the tyrosine catabolic pathway by NTBC corrected the altered gene expression in the livers of mice with targetdisrupted Fah and rescued the lethal phenotype (16). This altered gene expression seen in these mice (16) and in c 14CoS mice in the perinatal period (6 -11) is probably part of the cellular response to the apoptotic insults. Because mutant HPD apparently prevented both liver injury and carcinogenesis, the endogenous insults caused by the intracellular metabolite(s) may lead to apoptotic cell death or carcinogenesis in the livers of HT1 subjects. There are genetic diseases that predispose to cancer, such as defects in the repair pathway for DNA damage or increased sensitivity to chromosome breakage. An enhanced apoptosis has been demonstrated under experimental conditions in which DNA damage is increased or cell cycle regulation is impaired (30). These investigations support the notion that accumulation of DNA damage commits the cell to apoptosis or carcinogenesis (31). These DNA damages are caused either by endogenous insults produced by normal cellular metabolic events or by exogenous events (32). Our investigations, together with others on models and on HT1 patients, suggest that HT1 may be the first example of a genetic disease in which the accumulation of endogenous metabolite(s) due to a metabolic defect commits the cells to apoptosis or carcinogenesis.
Sudden apoptotic death of unmasked phenotype of HT1 in mature and unmodified hepatocytes had not been expected (1)(2)(3)(4)(5)(6)(7)(8)(9)(10)(11)(12)16), and these are implications for the pathogenesis and treatment of liver disease in HT1 patients. We suggest that mature and unmodified hepatocytes in those with the FAH defect cannot survive and that hepatocytes in the chronic form of HT1 have to be protected from a likely acute death. Both delayed development of the tyrosine catabolic pathway in the neonatal period and secondary inactivation of the pathway seem to contribute to survival of hepatocytes in HT1 patients. Indeed, HPD activities are reduced in the livers of HT1 patients (33), and this reduction is proposed to be part of the altered gene expression (34). If this inactivation is inadequate, acute death of hepatocytes is inevitable after the full expression of HPD, the result being the acute form of HT1. Under these circumstances, various cellular responses will occur as "adaptive dysregulation" or "adaptive mutation." The pleiomorphic appearance of the affected hepatocytes in the acute and severe forms of the disease (2) possibly reflects an unsuccessful process of survival of the fittest. These observations provide added support for the current prescription of NTBC for the treatment of HT1 patients (4).
Self-induced correction of mutations (revertant) on the Fah gene and proliferation of the corrected hepatocytes have been observed in the livers of chronic form HT1 patients (35), implying that hepatocytes corrected for FAH deficiency by gene transfer can expand in the HT1 liver. Following this observation, it was shown in mice carrying targeted-disrupted Fah that transplantation of hepatocytes that expressed FAH by retroviral mediated gene transfer under ex vivo conditions resulted in expansion of the cells in the liver (36). This provides an important model system in which one can investigate the fate of genetically corrected and transplanted hepatocytes in the liver where surrounding uncorrected cells were injured. The double mutant mice presented here will serve as another model for such experiments to investigate the fate of normalized hepatocytes in the damaged liver. In addition, both the double mutant mice and the target-disrupted FAH-deficient mice will be useful for the study of apoptosis of hepatocytes triggered by endogenous metabolite(s). Inhibition of the apoptosis of these hepatocytes with specific inhibitors will provide further understanding of hepatocyte injury and may provide alternative therapeutic interventions. Our preliminary exper-iments suggest that homogentisate-induced apoptosis is completely inhibited by certain specific inhibitors of apoptosis. In addition, the double mutant will provide an opportunity to examine the process of development of hepatocellular carcinomas caused by endogenous metabolite(s) in HT1.