Main

PKU (McKusick 261600) is an autosomal recessive disorder caused by deficiency of PAH (L-phenylalanine-4-monooxygenase; EC 1.14.16.1), which is predominantly expressed in the liver. PAH catalyzes the conversion of phenylalanine to tyrosine using tetrahydrobiopterin as a cofactor. The PAH gene was assigned to chromosome 12q22-24.1(1). Pathogenic mutations causing PKU have been extensively studied to date(2). Patients with PKU show profound mental retardation and hypopigmentation of skin, hair, and eyes due to increased amounts of phenylalanine in body fluids. Typically, the serum phenylalanine concentration exceeds 1200 µM (20 mg/dL; normal range, 40-110 µM). Mental retardation is often accompanied by other neurologic abnormalities, such as microcephaly, seizures, and hyperactive movements. If untreated, most patients lose about 50 points in IQ by the end of the 1st year of life and require institutional care(3). Although hypopigmentation of skin, hair, and eyes is a subtle clinical symptom among Caucasians, it is prominent among dark races such as the Japanese: PKU patients have fair skin, blond hair, and blue eyes, while their unaffected parents and siblings have dark skin, black hair, and brown eyes(4). These clinical symptoms can be prevented by a strict low-phenylalanine diet from early infancy. However, the rigid dietary therapy is a heavy burden on patients and their families, as it must not be relaxed before patients reach adolescence. Poor compliance often leads to unsatisfactory clinical outcome in PKU patients. Therefore, development of a safe and effective gene transfer method for the treatment of PKU has been long awaited(5). We believe that such gene therapy techniques will be useful for many other inborn errors of metabolism caused by hepatic enzyme deficiencies.

Gene therapy experiments in PKU have been reported previously. The attempts were facilitated by the development of a mouse model of PKU after treatment of inbred BTBR mice with N-ethyl-N-nitrosourea, an alkylating mutagen(6). Three mutant alleles were identified: Pahenu1, Pahenu2, and Pahenu3. Serum phenylalanine concentration in Pahenu2 or Pahenu3 was highly elevated on a regular diet. These mice showed behavioral abnormalities and pronounced hypopigmentation of the coat color similar to the clinical picture in PKU patients(6).

Early studies showed that a recombinant retrovirus vector was able to induce PAH activity in PAH-deficient cells in vitro(5). However, in vivo or ex vivo retrovirus-mediated gene transfer was not reported. Fang et al.(7) described adenovirus-mediated gene transfer into PKU mice. Recombinant adenovirus carrying human PAH cDNA under the control of the Rous sarcoma virus long terminal repeat was injected via the portal vein of PKU mice. Although a transient normalization of serum phenylalanine was observed after 1 wk, no additional phenotypic change was described. Therefore, it remains to be elucidated whether gene therapy can reverse pathologic symptoms in PKU mice, in addition to biochemical parameters. Also, in the latter study, a second viral challenge failed to lower the phenylalanine level, probably due to viral clearance by neutralizing antibody. The short duration of therapeutic effect without the possibility of repeated gene transfer must be circumvented if this technique is to be considered for clinical application, inasmuch as serum phenylalanine in PKU must be maintained at a low level for a considerable length of time.

In this study, we attempted to produce more sustained therapeutic effects of gene transfer in PKU mice by using a recombinant adenovirus harboring human PAH cDNA under the control of a CAG promoter, one of the most potent promoters available today, with the aid of the immunosuppressant, FK506, to reduce the immunologic response of recipient mice. We administered the virus particles to PKU mice to determine whether gene therapy could reverse pathologic phenotypes in these mice. In these experiments, two different routes of administration of the virus, the portal vein versus the tail vein, were compared to explore the possibility of in vivo gene therapy using a less invasive approach. FK506 markedly suppressed the host immune reaction as demonstrated by biochemical, histologic, and immunologic analyses and allowed prolonged expression of the PAH gene and repeated gene delivery.

METHODS

PKU mouse model. A strain of PAH-deficient mice, Pahenu2, was a generous gift of Dr. Alexandra Shedlovsky(6). Clinical features of the mouse model resembled those of human PKU patients: hypopigmentation, behavioral abnormalities, and a 10- to 20-fold elevation in serum phenylalanine level (1236 ± 366 µM, range: 768-2268, n = 39). Hepatic PAH activity was undetectable; Western blot analysis showed a reduced amount of cross-reacting material in the liver, and PAH mRNA was decreased to less than 1% of that in wild-type liver tissue(6). Molecular analysis identified a T-to-C transition at position 835 in PAH cDNA. This mutation resulted in the substitution of a phenylalanine residue with serine at amino acid position 263 of the mouse enzyme(8). Pups born to homozygous females did not survive beyond several hours. Because of this maternal effect, heterozygous females were bred with homozygous males to obtain homozygous offspring. Throughout the experiments, mice had unlimited access to water and a diet that contained 21.3% protein (0.89% phenylalanine; F-1, Funahashi Farms, Funahashi, Japan).

Human PAH cDNA. We screened a human liver cDNA library to obtain full-length PAH cDNA. The isolated PAH cDNA clone contained the entire coding region and carried no base substitution according to sequencing analysis using an A.L.F. automated laser fluorescent sequencer (Pharmacia Biotech, Uppsala, Sweden). When the PAH cDNA was cloned into an eukaryotic expression vector pUC-CAGGS(9) and introduced to COS1 cells, a marked increase in PAH activity was observed (data not shown).

Construction of recombinant adenovirus. Recombinant adenovirus vector containing human PAH cDNA was constructed by a previously described COS-TPC method(10). Briefly, human PAH cDNA was excised from the expression vector pUC-CAGGS-hPAH by EcoR I digestion, blunt-ended with a Klenow fragment of DNA polymerase I, and ligated with an Swa I-digested cosmid cassette vector pAxCAwt(11). The pAxCAwt contained the whole adenovirus genome except E1A, E1B, and E3. The direction of the inserted unit was analyzed by Xho I-digestion, and clones that had PAH cDNA downstream of the CAG promoter in a "sense" orientation were selected. The CAG promoter consisted of the cytomegalovirus immediate early enhancer and a chicken β-actin/rabbit β-globin hybrid promoter(9). The expression unit also contained rabbit β-globin 3′-flanking sequences and a polyadenylation signal. We used Ad5-dlX(12) as a parent virus for recombinant adenovirus construction. The parent adenovirus DNA-terminal protein complex (Ad-DNA-TPC) was digested with EcoT22 I and purified by gel filtration. Eight micrograms of the cosmid cassette was mixed with 1 µg of the digested Ad-DNA-TPC and used to transfect 293 cells in a 6-cm dish by the calcium-phosphate precipitation method using a CellPhect Transfection kit (Pharmacia Biotech). The desired recombinant adenovirus was generated through homologous recombination between the common regions of Ad5-dlX and the cosmid cassette in 293 cells. After 24 h, the cells were spread in three 96-well plates at a 10-fold serial dilution mixed with untransfected 293 cells which supplied gene products from E1A and E1B. The culture was maintained for 10-15 d, and recombinant viral clones were obtained from wells in which cytopathic effects were observed due to the propagation of recombinant viruses.

Isolated viral clones were propagated by a standard procedure, purified by two rounds of centrifugation in a CsCl density gradient, dialyzed in PBS with 10% glycerol, and stored at -80°C(13).

Titration of recombinant adenovirus. Fifty microliters of DMEM supplemented with 5% FBS was dispensed into each well of a 96-well tissue culture plate coated with rat type I collagen. Eight rows of 3-fold serial dilution of the virus starting from a 10-4 dilution were prepared on the plate. A total of 3 × 105 293 cells in 50 µL of DMEM/5% FBS was added to each well, and the plate was incubated at 37°C in 5% CO2 in air. Fifty microliters of DMEM/5% FBS was added to each well every 3 d. On d 12, the endpoint of the cytopathic effect was determined by microscopy, and the 50% tissue culture infection dose was calculated according to Kärber's method(13).

Transfection of COS7 cells in vitro with the recombinant adenovirus. COS7 cells were cultured in minimum essential medium containing 10% FBS, 20 mM glucose, 100 U/mL penicillin, and 100 µg/mL streptomycin and grown to confluence at 37°C in 5% CO2 in humidified air. After the culture medium was removed, cells were exposed to various titers of recombinant adenovirus for 1 h. The cells were further cultured and harvested after 24, 48, and 78 h for measuring PAH activity.

Administration of recombinant adenovirus to PKU mice in vivo. All animal experiments were conducted in accordance with the Guidelines for Animal Experiments at Tohoku University. PKU mice used for in vivo gene transfer were 10 wk of age. The recombinant adenovirus was administered to mice via the portal vein or tail vein. For portal vein infusion, mice were anesthetized with ether inhalation followed by laparotomy. A solution containing 3.9 × 106 to 3.9 × 108 pfu of the recombinant virus in 100 µL of saline was slowly injected into the portal vein using a 30-gauge needle. Control mice received normal saline. The appearance and activity of each mouse were carefully observed, and body weight was measured daily. Tail vein infusion was performed with 100 µL of solution containing 1.2 × 108 to 1.2 × 109 pfu of the virus using an insulin infusion syringe equipped with a 29-gauge needle (Becton-Dickinson, Lincoln Park, NJ).

Heparinized capillary vessels were used to periodically collect blood samples from tail veins. For the measurement of serum phenylalanine, 40 µL of whole blood was absorbed on filter paper originally developed for newborn mass screening, dried, and stored at room temperature until analysis. Serum was also separated for the determination of transaminase levels to monitor liver function. On the indicated days, mice were killed with an overdose of inhaled ether, and tissues were resected and stored at -80°C for PAH assay, PCR analysis, and pathologic examinations. Skin biopsy was also performed.

FK506 treatment. FK506 (tacrolimus) was generously provided by Fujisawa Pharmaceutical Co. (Osaka, Japan). In immunosuppression experiments, PKU mice received daily s.c. injection of FK506 at a dose of 5 mg/kg, beginning on the day of gene delivery.

Serum phenylalanine measurement. Serum phenylalanine was determined by an Enzaplate N-PKU kit (Chiron, Emeryville, CA). Briefly, a disc (3 mm in diameter) was punched out from a dried blood spot and placed in a 96-well microtiter plate. Phenylalanine was eluted from the disc and incubated with phenylalanine dehydrogenase, which catalyzed the NAD-dependent deamination of phenylalanine to phenylpyruvate and ammonia, thus reducing NAD to NADH. The resulting NADH served as the electron donor in a colorimetric detection system composed of cobalt (III) acetylacetonate/1-methoxy-5-methylphenazium methylsulfate/2-(5-bromo-2-pyridyazo)-5-(N-propyl-N-sulfopropylamino) phenol(14). The absorbance at 590 nm was measured by a microplate reader, model 450 (Bio-Rad, Hercules, CA).

PAH assay. Measurement of PAH activity in tissues was based on a previously reported method(15). The frozen tissue was homogenized in 0.15 M KCl, sonicated, and centrifuged, and the supernatant was used for PAH assay with [14C]phenylalanine (Amersham, Buckinghamshire, UK) as a substrate. After incubation at 25°C for 20 min, the deproteinized reaction mixture was applied onto a thin-layer chromatography system (TLC aluminum sheets silica gel 60, Merck, Darmstadt, Germany) and developed for 90 min at room temperature in CHCl3-CH3OH-NH4OH-H2O (58:32:8:2, vol/vol). The plate was sprayed with a ninhydrin solution so that the location of phenylalanine and tyrosine could be visualized. Each amino acid spot was cut out with scissors, placed in a scintillation vial containing liquid scintillant, and measured for radioactivity of [14C]phenylalanine and converted [14C]tyrosine.

Detection of recombinant adenoviral DNA in tissues by PCR. DNA was isolated from mouse organs by a proteinase K digestion/phenol extraction/ethanol precipitation method. The viral genomic DNA was detected by PCR amplification using the following primers: 5′-GGTTGTTGTGCTGTCTCATC-3′(AD/HPAH-F1) and 5′-GCCAATGCACCAACTTCTTC-3′(AD/HPAH-R1). AD/HPAH-F1 was complementary to a region in the CAG promoter, and AD/HPAH-R1 was located within the PAH cDNA sequence. The size of the targeted sequence was 255 bp. The thermoprofile for PCR reaction included initial denaturation for 3 min at 94°C followed by 30 cycles of denaturation at 94°C for 30 sec, annealing at 55°C for 30 sec and extension at 30°C for 30 sec.

The amplified DNA fragment was electrophoresed in a 3% agarose gel, stained with ethidium bromide and illuminated by UV light. The visualized DNA bands were recorded with a CCD camera, and the intensity of each band was measured by a DensitoGraph Version 1.0 (Atto, Tokyo, Japan) to calculate the amount of viral DNA in each tissue in comparison with those obtained by amplification of known titers of recombinant adenoviruses.

Histologic examination. The liver and skin tissues were fixed in 10% formaldehyde, embedded in paraffin, sectioned, and stained with hematoxylin and eosin.

Titration of neutralizing antibodies against adenovirus. Adenovirus neutralization assay was performed according to a previously described method(16). Briefly, serum samples taken from mice were incubated for 30 min at 56°C to inactivate complement, serially diluted with heat-inactivated serum from untreated mice, and incubated with 1 × 106 pfu of recombinant adenovirus for 1 h at room temperature. The mixture was used to infect 293 cells cultured in a 96-well plate coated with type I collagen. Neutralizing antibody titers were determined by the highest dilution of serum at which cell viability was more than 50% after 4 d of infection.

Liver function tests. ALT was assayed by a Transaminase-CII-Test kit (Wako, Tokyo, Japan).

RESULTS

Vector construction. A replication-defective recombinant adenovirus was constructed by homologous recombination between the cosmid-cassette vector harboring human PAH cDNA and EcoT22 I-digested Adex5dlX in 293 cells. Nine viral clones were selected after three rounds of screening. To confirm the integration of human PAH cDNA in a targeted site of the adenovirus vector, the viral DNA was digested by Xho I. Electrophoresis of Xho I-digested DNA fragments indicated that the desired integration of human PAH cDNA occurred in all recombinants (data not shown). The viruses were devoid of the E1A, E1B, and E3 regions and carried a PAH expression unit at the E1-deleted region. The transcriptional orientation of the expression unit was opposite to the original orientation of the E1A and E1B genes.

In vitro PAH expression in COS7 cells by the recombinant adenovirus. Infection of COS7 cells in vitro with each of the isolated recombinant viruses produced significant PAH activity. When one of the viral clones (Adex1CA-Y-hPAH17) was used to infect at a multiplicity of infection of 1.0, the enzymatic activity increased to approximately 30% of that in normal hepatocytes (84.9 ± 6.0 nmol of tyrosine formed per milligram of protein per hour, n = 4) after 24 h of gene transfer. The increase in PAH activity took place in a dose-dependent manner over a range of multiplicity of infection of 0.1 to 3.0.

In vivo adenovirus-mediated gene transfer in PKU mice. First, Adex1CA-Y-hPAH17 was infused into the liver of PKU mice via the portal vein at various doses. Serum phenylalanine was then monitored at 12 h and 1, 3, 5, 7, 11, and 14 d. Serum phenylalanine in mice (n = 3) that received more than 7.5 × 107 pfu of the virus dramatically decreased from 1014 ± 114 µM to 150 ± 36 µM within 24 h. The phenylalanine concentration remained low for 7 d and then increased to greater than 1200 µM on d 10. In contrast, mice infused with 3.9 × 106 pfu of viruses showed no reduction in serum phenylalanine levels.

Next, we delivered the recombinant adenovirus (1.2 × 108 to 1.2 × .109 pfu/mouse) into the systemic circulation via the tail vein. When the virus was administered at doses of greater than 3.6 × 108 pfu/mouse, the high serum phenylalanine concentration (1260 ± 330 µM, n = 11) decreased to 96 ± 48 µM (Fig. 1) and remained in the normal range for 7 d. A higher viral dose of 1.2 × 109 pfu/mouse significantly increased the duration of reduced serum phenylalanine (7 d versus 11 d, p < 0.01). However, a second viral challenge on d 50 or d 110 failed to reduce the serum phenylalanine concentration.

Figure 1
figure 1

Serum phenylalanine concentration after tail vein infusion of Adex1CA-Y-hPAH17 in PKU mice. Ten-week-old PKU mice were injected with 100 µL of saline (open circles, n = 3) or the recombinant virus (closed squares, 1.2 × 108 pfu, n = 4; closed triangles, 3.6 × 108 pfu, n = 3; closed circles, 1.2 × 109 pfu, n = 3) via the tail vein.

Full size image

An additional prominent change was observed in treated PKU mice. Hypopigmentation of the coat was gradually converted from a grayish color to black. This effect was first observed on d 7-10 around the eyes and extended to the whole body after 2 wk. At this point it was impossible to distinguish PKU mice from wild-type mice. However, as the phenylalanine level increased again, pigmentation was gradually diminished until the coat color returned to gray in 2 wk. These effects on pigmentation were correlated with those on serum phenylalanine levels: low-dose gene transfer that did not alter phenylalanine concentration failed to induce pigmentation. The duration of pigmentation was significantly prolonged by FK506 administration (see below).

Tissue distribution of recombinant adenovirus and PAH activity after gene therapy administered via the tail vein. Tissue distributions of recombinant adenovirus and PAH activity were examined after 3 d of gene transfer in PKU mice (n = 4) infused with viral doses of 4.6 × 108 pfu/mouse. The amount of viral DNA was semiquantified by PCR amplification. The majority of the recombinant virus was found in the liver (1.2 × 104 pfu/µg of tissue DNA; Fig. 2). Viral DNA was also detected, in much lower quantities, in the rectum (8.1 × 102 pfu/µg of tissue DNA), quadriceps femoris muscle (4.4 × 102 pfu/µg of tissue DNA), spleen (3.5 × 102 pfu/µg of tissue DNA), and kidney (1.6 × 102 pfu/µg of tissue DNA) and was undetectable in brain, heart, lung, or testis. PAH activity was observed exclusively in the liver (44.6 ± 16.8% of the PAH activity in normal hepatocytes, n = 4; Fig. 2). As little as 9% of the normal PAH activity in the liver was sufficient to maintain serum phenylalanine levels in the normal range (data not shown).

Figure 2
figure 2

Tissue distribution of the recombinant adenoviral DNA after tail vein administration. The PAH activity (A) and the amount of adenovirus vector DNA (B) were measured in brain, heart, lung, liver, kidney, rectum, testis, muscle, and spleen. Specimens were obtained 3 d after tail vein injection of Adex1CA-Y-hPAH17 (4.6 × 108 pfu). The PAH activity per milligram of protein in each tissue was compared with the reference value of 100% in wild-type hepatocytes.

Full size image

Liver function and histologic examination of liver after gene therapy. On d 5 of the gene delivery via the tail vein, we noticed icteric serum in which ALT was significantly elevated. The values peaked on d 7 (ALT = 696 ± 162.5 KU [range: 534.2-928.7, n = 5]; normal: 26.4 ± 5.03) and decreased to normal levels 2 wk after infection. The increase coincided with histologic findings of acute hepatitis (Fig. 3A) on d 7. Marked infiltration of mononuclear inflammatory cells in sinusoids associated with disrupted alignment of hepatocytes was also observed. The liver also showed formation of acidophilic bodies and swollen Kupffer cells.

Figure 3
figure 3

Histology of the liver after gene therapy with Adex1CA-Y-hPAH17. Mice were treated with daily s.c. injections of 5 mg/kg of FK50 6 (B) or were not given FK506 (A). Hematoxylin and eosin were used to stain sections obtained 7 d after tail vein injection of 1.2 × 109 pfu of Adex1CA-Y-hPAH17. Original magnification: 100×.

Full size image

Immunosuppression by FK506. To limit the host immune response associated with adenovirus-mediated gene transfer, we treated mice with daily s.c. injection of FK506. FK506 has been shown to block signal transduction of cytokines, with consequent strong immunosuppression. When PKU mice received 1.2 × 109 pfu of the adenovirus i.v. along with FK506, the serum phenylalanine level decreased in 24 h to normal levels and remained significantly different from that in control mice for 47 d (p < 0.05) (Fig. 4). This result is in sharp contrast to findings in mice that had undergone gene therapy without FK506 (Fig. 4). The prolongation of the gene therapy effect coincided with the absence of liver dysfunction as determined both by ALT values on d 7 (44.8 ± 19.9 KU, range: 26.8-80.8, n = 6) and histologic examination (Fig. 3B). Furthermore, when a second recombinant adenovirus challenge was administered 110 d after the first, the serum phenylalanine concentration was again reduced in mice treated with FK506 (Fig. 4). The duration of the normal serum phenylalanine level was, however, significantly shorter than that observed after the first administration (11.3 d versus 35.8 d, p < 0.01).

Figure 4
figure 4

Effect of FK506 on adenovirus-mediated gene transfer. Mice were infused with 1.2 × 109 pfu of Adex1CA-Y-hPAH17 through the tail vein on d 0 and treated with 5 mg/kg/d of FK506 (closed circles, n = 3) or no FK506 (open circles, n = 3). On d 110, the mice received a second administration of the recombinant virus at the same dose.

Full size image

The effect of FK506 on the production of neutralizing antibodies against the recombinant virus was studied in a second series of gene transfer experiments. No mice had neutralizing antibodies before gene transfer (Fig. 5). Mice without FK506 (n = 3) showed reduced phenylalanine levels for 10 d and developed neutralizing antibodies with a titer of 1:10-1:100 at 1 wk after gene transfer with recombinant viral doses of 4.6 × 108 pfu/mouse. This antibody titer persisted until d 55, when a second viral infusion was performed. The second challenge failed to alter the phenylalanine concentration and was followed by a marked increase in the antibody titer (1:300-1:600). In contrast, mice treated with FK506 (n = 3) did not produce detectable amounts of antibodies despite repeated gene delivery. One mouse, however, developed antibodies with a titer of 1:100 on d 42 despite FK506 treatment (Fig. 5); this mouse responded poorly to the second administration of the virus.

Figure 5
figure 5

Neutralizing antibodies against adenovirus in serum from mice that received Adex1CA-Y-hPAH17 on d 0 and 55 at a dose of 4.6 × 108 pfu/mouse. Mice were treated with daily s.c. injections of 5 mg/kg of FK506 (closed circles, n = 3) or no FK506 (open circles, n = 3). On d 0, blood samples were collected before the infusion of adenoviruses.

Full size image

The effect of FK506 on PAH activity in the liver was examined on d 3: mice treated with FK506 showed greater PAH activity than nontreated mice. This difference was also observed on d 21, when PAH activity was no longer detected in nontreated mice (data not shown).

The reversal of hypopigmentation in FK506-treated mice after gene transfer was essentially the same as that observed in non-FK506-treated mice, except that this response lasted for more than 40 d (Fig. 6B, right). Initial periorbital pigmentation on d 7 and a gradual decrease in pigmented area after serum phenylalanine elevation were also observed, as in non-FK506-treated mice (Fig. 6,A and C).

Figure 6
figure 6

Changes in coat color in PKU mice after gene therapy. A mouse injected with Adex1CA-Y-hPAH17 showed periorbital pigmentation, an initial sign of coat color change, on d 7 (A). The mouse shown in B, right, exhibited complete coat color change at 35 d after gene transfer with daily injection of FK506. An untreated mouse is shown on the left for comparison. C shows a mouse at 35 d after adenovirus infection without FK506 treatment: the pigmented region diminished to a limited area on the back.

Full size image

Histologic examination of biopsied skin. Hematoxylin-eosin-stained skin specimens from an untreated PKU mouse revealed poorly grown hair follicles located in the dermis (Fig. 7A). Hairs were hypopigmented and a few melanin granules were observed in the hair follicles. In contrast, after the correction of hyperphenylalaninemia for 35 d by gene transfer, hair follicles were deeply situated in the s.c. adipose tissue (Fig. 7B), and more melanin granules were observed in both hairs and follicles.

Figure 7
figure 7

Histology of skin and hair follicles. Hematoxylin- and eosin-stained sections of skin were obtained before gene transduction (A) and 35 d after gene transduction (B). Original magnification: 10×.

Full size image

DISCUSSION

The most striking observation in the current study was the complete, albeit transient, reversal of hypopigmentation in PKU mice (Pahenu2) after the adenovirus-mediated introduction of PAH cDNA. It has been reported that when PKU mice were placed on a low-phenylalanine diet to maintain a low serum phenylalanine level, hypopigmentation was partially reversed(6). This effect is similar to that observed in PKU patients on low-protein diets(17), indicating that the correction of hypopigmentation is an important marker for evaluating the efficacy of gene therapy for PKU. In the previous adenovirus-mediated gene transfer experiments in Pahenu2 mice by Fang et al., however, no reversal of hypopigmentation was described in the original report or subsequent articles(7,18,19). In that study, serum phenylalanine concentrations were measured at 1-wk intervals after the injection of 10 × 1010 pfu of viral particles per mouse via the portal vein; a reduction in phenylalanine levels was observed 1 wk after gene delivery, but this value increased to the pretreatment level after 2 wk(7). The biochemical changes were nearly equivalent to those in our experiments using viral doses of 7.5 × 107 pfu/mouse injected through the portal vein (without FK506 treatment), in which the normalization of serum phenylalanine level for 7-10 d was sufficient to cause a distinct change in coat color.

Hypopigmentation in PKU patients has been attributed to impaired tyrosine metabolism due to hyperphenylalaninemia. The hydroxylation of tyrosine, catalyzed by tyrosinase, is the first step in the formation of melanin pigment. Tyrosinase is known to be competitively inhibited by increased amounts of L-phenylalanine, thus reducing melanin production(20). In addition, phenylalanine competitively inhibited tyrosine uptake by melanocytes in a melanoma cell line(21). Therefore, the reversal of hypopigmentation in PKU mice appears to be a direct consequence of lowered phenylalanine concentrations in body fluids after gene therapy. Electron microscopic analysis of the biopsied skin showed increased production of premelanosomes and melanosomes in melanocytes after gene therapy (data not shown).

The increased pigmentation of the coat after gene therapy was associated with a decrease in hair loss. The prominent loss of hair in untreated PKU mice seems to be the result of an unrelated, recessive, tufted (tf) mutation carried in the background BTBR strain. However, gene therapy appeared to have a modifying effect on this phenotype: after gene transfer, hairs became more firmly anchored in the skin and were not easily dislodged. Our observations might correspond to dermatologic changes observed in PKU patients, such as eczematous dermatitis and scleroderma-like skin lesions(4,22). These symptoms respond well to dietary restriction of phenylalanine, suggesting an adverse influence of hyperphenylalaninemia on skin metabolism.

The newly constructed vector, Adex1CA-Y-hPAH17, led to normalization of serum phenylalanine for 7-10 d at a dose of 7.5 × 107 pfu via the portal vein or 3.6 × 108 pfu via the tail vein. The apparently strong transgene expression in vivo was probably due to the use of the CAG promoter to drive PAH cDNA expression. In an experiment using a primary human hepatocyte culture, the CAG promoter exhibited 10-40 times higher gene expression than the SRα promoter(16). The use of a potent promoter in our study permitted a reduction in the titer of recombinant virus that is required to obtain critical transgene expression. In Figure 1, a transient elevation of the serum phenylalanine levels of mice treated with 1.2 × 108 pfu was observed after d 7. This may be explained by accelerated protein catabolism due to viral infection and liver dysfunction caused by host immune response. At higher viral doses, these responses were probably masked by increased PAH activities.

Because the PAH gene is exclusively expressed in the liver in humans, hepatic tissue was targeted in our experiment. We first infused adenoviruses into the liver through the portal vein of PKU mice to obtain high levels of expression in liver tissue(7). However, this approach required an invasive surgical procedure, which might hamper the clinical application of this therapy, especially since the transient gene expression requires repeated vector administration. Systemic delivery via the peripheral tail vein resulted in transgene expression predominantly in the liver. The high affinity of adenoviral vectors for liver tissue is in line with previous reports, although the administration route of the virus appeared to influence the tissue distribution(23,24).

The therapeutic effects of gene transfer without FK506 lasted for only 7-10 d. Reelevation of the serum phenylalanine level was preceded by acute hepatitis associated with marked mononuclear cell infiltration and pronounced elevation of serum transaminases. Liver dysfunction was not observed during the initial 4 d after gene delivery, whereas transgene expression was already observed within 12 h. Therefore, hepatitis was likely to be caused by the immune response of the host rather than direct toxicity associated with adenovirus infection. This is in line with a previous report that the transient expression associated with adenovirus-mediated gene transfer can be caused by cellular immune responses of the host against adenovirus vectors(25). It is interesting that our preliminary experiments showed that recombinant (E1/E3-deleted) adenovirus without PAH cDNA was less immunogenic than Adex1CA-Y-hPAH17, suggesting that a transspecies immune response to human PAH protein is also responsible (data not shown).

To overcome the unfavorable immune responses, we administered an immunosuppressant, FK506, to the mice receiving gene transfer. Similar strategies have been explored to suppress the host immune response using various immunomodulators such as IL-12, CTLA4Ig, cyclosporin A, cyclophosphamide, deoxyspergualin, and FK506(26–31). FK506 is a macrolide antibiotic that inhibits T cell activation(32,33). It interferes with the first route signal transduction between T cell receptor and nucleus, thus suppressing translation activity of cytokines such as IL-2. FK506 is also known to interfere with the induction of nitric oxide synthesis in macrophages, which play a critical role in triggering immune responses. Application of FK506 to adenovirus-mediated gene transfer aimed at mouse skeletal muscles has been reported previously(30,31); the duration of foreign gene expression was prolonged by using FK506, whereas the level of transduction after a second administration was markedly lower than that achieved by the initial administration. Inasmuch as cyclosporin A, which has the same immunosuppressive mechanism as FK506, totally failed to allow transgene expression after repeated gene therapy(28), these studies suggested that this type of immunosuppressant is incompetent for inhibiting neutralizing antibody production. In contrast, our liver-directed gene therapy showed that the administration of FK506 not only prolonged PAH expression but also enabled a repeated gene transfer. The second round of gene transfer was, however, less effective than that observed after the first administration, suggesting that the immunosuppression by FK506 may not be complete. It should be noted that FK506 fully prevented the development of severe hepatitis, which would be a serious problem in future clinical applications. Recently, less immunogenic recombinant adenoviral vectors were developed, such as E1/E4-deleted or "gutted" vector, which lacks all viral genes(34,35). These new-generation vectors show reduced immunogenicity, offering considerable promise for long-term transgene expression by adenoviral gene transfer. However, even the "gutted" vector is not completely free of immune responses, because it still has immunogenic viral capsids provided by helper viruses during in vitro propagation. Therefore, it appears that the immunologic problems associated with adenovirus-mediated gene transfer cannot be solved by the improvement of vector design alone and that immunomodulation may be a requisite for effective and safe gene therapy.

In summary, we were able to reverse hypopigmentation, one of the major clinical phenotypes in PKU mice, by adenovirus-mediated gene transfer. These observations indicate the feasibility of gene therapy for PKU. It would be of interest to study the effects of gene therapy on neurologic impairment in PKU mice. We noticed that mice with normalized serum phenylalanine levels after gene therapy were more alert than their untreated counterparts, although the phenotypic changes remain to be elucidated by carefully designed behavioral studies. It has been reported that PKU mouse brain shows increased turnover of myelin without compensation by an increased rate of synthesis, thus leading to defective myelination and loss of neurotransmitter receptors(36,37). Also, studies in experimental rat models with chemically induced hyperphenylalaninemia revealed impaired performance on a cognitive task dependent on brain frontal cortex associated with reduced homovanillic acid levels in the same region(38). Evaluation of such neurologic parameters in PKU mice after adenovirus-mediated gene transfer could provide further experimental evidence for the efficacy of gene therapy in PKU.