Complete and persistent phenotypic correction of phenylketonuria in mice by site-specific genome integration of murine phenylalanine hydroxylase cDNA

Abstract

We explored the potential of using a bacteriophage integrase system to achieve site-specific genome integration of murine phenylalanine hydroxylase cDNA in the livers of phenylketonuric (PKU) mice. The phiBT1 phage integrase is an enzyme that catalyses the efficient recombination between unique sequences in the phage and bacterial genomes, leading to the site-specific integration of the former into the latter in a unidirectional manner. Here we showed that this phage integrase functions efficiently in mouse cells, and several naturally occurring pseudo-attP sites located in the intergenic regions of the mouse genome have been identified and molecularly characterized. We further demonstrated that the addition of nuclear localization signal sequences to the C terminus of the phage integrase enhanced the efficiency for transgene integration into the mouse genome. Using this phage integration system, we delivered mouse phenylalanine hydroxylase cDNA to the livers of PKU mice by hydrodynamic injection of plasmid DNA and showed that the severity of the hyperphenylalaninemic phenotype in the treated mice decreased significantly. After three applications, serum phenylalanine levels in all treated PKU mice were reduced to the normal range and remained stable thereafter. Their fur color also changed from gray to black, indicating the reconstitution of melanin biosynthesis as a result of available tyrosine derived from reconstituted phenylalanine hydroxylation in the liver. Thus, the phiBT1 bacteriophage integrase represents an effective site-specific genome integration system in mammalian cells and can be of great value in DNA-mediated gene therapy for a multitude of genetic disorders.

Fig. 1.

PhiBT1 integration system. (A) Schematic representation of site-specific phage DNA integration into host genome. The phage integrase recognizes the attB and attP sites in the bacterial and phage genomes, respectively, as substrates and catalyzes the recombination reaction between them, resulting in the integration of phage DNA into the host genome. This is a unidirectional reaction because the products of recombination, attL and attR, are not recognized by the integrase. In nature, the integrated phage DNA can only be rescued from the bacterial genome by action of excisionase, which is a distinct phage-encoded enzyme that recognizes attL and attR sequences as substrates for recombination. (B) Schematic maps of plasmids. Plac, lacZ promoter; chl, chloramphenicol resistant gene; int, phiBT1 integrase gene; kan, kanamycin resistant gene; CMV, CMV promoter; NLS, SV40 nuclear localization signal; CAG, CAG promoter; IRES, the internal ribosome entry site from the encephalomyocarditis virus (ECMV); SEAP, reporter gene secreted alkaline phosphatase; mPAH, mouse PAH gene cDNA. (C) SEAP expression in transfected mouse 3T3 cells after serial passaging. Mouse 3T3 cells were cotransfected with the integrase-expressing plasmids (pCMV-BTInt or pCMV-BTIntNLS) and reporter plasmids without (pCZiS) or with (pCZiS-B) an attB sequence at the ratio of 20:1. SEAP concentrations in conditioned media of all groups rose to similar levels after transfection, which returned to background in the attB-negative group after serial passaging. Those in the attB-positive groups, however, remained stable at a reduced level after multiple cell passages (P < 0.05 for normal integrase group, and P < 0.01 for NLS-containing integrase group). Additionally, the presence of SV40NLS sequence in the integrase enzyme resulted in a 4-fold higher level of SEAP expression in the transfected cells after serial passaging (P < 0.05).

Fig. 2.

Characterization of the pseudo-attP sites in the mouse genome. (A) Integrated plasmids were rescued from total genomic DNA of transfected mouse 3T3 cells by bacterial cloning, and the flanking mouse DNA sequences were determined. The flanking mouse DNA sequences were compared with the available mouse genome sequence, and the corresponding phage integration sites were determined. These sites represent chromosomal locations of the pseudo-attP sties naturally occurring in the mouse genome. Eight different pseudo-attP sites were identified, which are located in mouse chromosomes 3, 5, 7, 4, 10, 9, and 8, respectively. All of these sites are located within the intergenic regions of the respective chromosomes. (B) Multiple alignments of the pseudo-attP sites in mouse genome. Compared with the wild-type attP sequence, the pseudo-attP showed only 35–50% sequence homology. The conservative regions are shown as shaded areas. (C) Integration frequency in vitro. To quantify the frequency of integration events at each pseudo-attP site, primers were designed to amplify the attL junctions. Standard dilution curves were established by using control plasmid DNAs of known concentrations. The data were normalized to genome copy number by using primers that amplify the single-copy mouse PAH gene per haploid genome. In 3T3 cells, >76% of integration events occurred at the major site mpsP3 (0.013 per haploid genome) and another 23% occurred in the minor sites mpsP5 and mpsP7 (0.0024 and 0.0017 per haploid genome). The integration frequencies on the other minor pseudo-attP sites were close to the detection limit.

Fig. 3.

Site-specific genome integration in vivo.(A) Kinetic profiles of SEAP in treated mouse sera. Normal CD1 mice were injected with the integrating plasmid containing the SEAP expression cassette and buffer, plasmids expressing the normal integrase, or NLS-modified integrase. In the normal integrase treatment group (open circles), SEAP expression reached its peak at 24 h postinjection, then decreased in 2 weeks and remained stable at a low level. The group treated with modified integrase (filled squares) had higher levels of SEAP in their serum samples (P < 0.05). In the integrase-negative group (filled triangles), no SEAP expression was detected at 2 weeks postinjection. (B) In vivo integration frequency. The genomic DNA samples from the livers of treated mice were used for real-time PCR assays to determine the integration frequencies at various pseudo-attP sites. More than 95% of integration events occurred at the major site mpsP3 (1.762% per haploid genome). Another 4% of integration events occurred in the two minor sites mpsP5 and mpsP7 (0.048% and 0.044% per haploid genome). Less than 1% of integration events occurred in the other minor sites.

Fig. 4.

Phenotype correction of severe hyperphenylalaninemia in PKU mice after treatment with an integrating plasmid expressing mouse PAH cDNA. (A) Plasmid map of pCmPAH-B. The vector contains a mouse PAH gene cDNA expression cassette driven by the CAG promoter (31) and a wild-type attB sequence. (B) Serum phenylalanine curves of treated PKU mice. Each mouse received injections of either PBS or pCmPAH-B at three different time points (shown with arrow). Serum phenylalanine levels decreased abruptly in all groups (n = 6 per group) after each PAH vector administration. In the integrase-negative treatment group, serum phenylalanine concentrations returned to pretreatment levels (P > 0.1) within 2 weeks. In the integrase-positive treatment group, serum phenylalanine concentrations remained stable at a substantially reduced level (P < 0.01). After three administrations of the vector, serum phenylalanine levels in the integrase-positive treatment group decreased to the normal range (P > 0.1 when compared with that of normal mice) and remained at those levels thereafter. (C) Fur color change in PKU mice after treatment. After three consecutive administrations of the integrating plasmid DNA vector, no mice changed fur color in the integrase-negative control treatment group (Upper), whereas all PKU mice in the vector-treatment group (Lower) turned black. (D) PAH activities in liver extracts of normal and treated PKU mice. Specific enzymatic activities were defined as total cpm in [¹⁴C]tyrosine produced in the assay per mg of protein in the liver extracts, and the initial slopes of the saturation curves were compared statistically by the Student t test (P < 0.01). The cpm/mg protein values were calculated from the tangents of the curves. Each groups contained six animals. In treated PKU mice the total enzymatic activity was ≈17% of normal, whereas in the liver of heterozygous mice the level was 50% of normal as expected. In untreated PKU mice, the level was <1% of normal.