Copyright 0 1993 by the Genetics Society of America

Mouse Models of Human Phenylketonuria

Alexandra Shedlovsky, J. David McDonald,' Derek Symula and William F. Dove McArdle Laboratory for Cancer Research and Laboratory of Genetics, University of Wisconsin, Madison, Wisconsin 53706

Manuscript received January 2 1, 1993 Accepted for publication April 29, 1993

ABSTRACT Phenylketonuria (PKU) results from a deficiency in phenylalanine hydroxylase, the enzyme catalyz-

ing the conversion of phenylalanine (PHE) to tyrosine. Although this inborn error of metabolism was among the first in humans to be understood biochemically and genetically, little is known of the mechanism(s) involved in the pathology of PKU. We have combined mouse germline mutagenesis with screens for hyperphenylalaninemia to isolate three mutants deficient in phenylalanine hydroxylase (PAH) activity and cross-reactive protein. Two of these have reduced PAH mRNA and display characteristics of untreated human PKU patients. A low PHE diet partially reverses these abnormal- ities. Our success in using high frequency random germline point mutagenesis to obtain appropriate disease models illustrates how such mutagenesis can complement the emergent power of targeted mutagenesis in the mouse. The mutants now can be used as models in studying both maternal PKU - and somatic gene therapy.

P HENYLKETONURIA (PKU) in humans results from a deficiency in phenylalanine hydroxylase

(PAH) (SCRIVER and CLOW 1980). When untreated, this condition results in severe and irreversible mental retardation. Treatment consists of careful regulation of essential dietary phenylalanine (PHE) sufficient to provide for protein synthesis without excess accumu- lation. With precise dietary control, initiated shortly after birth, the detrimental effects of PKU can largely be avoided. Since these effects most severely impact the developing central nervous system, PKU patients have been weaned from the PHE-controlled diet in later childhood, although recent evidence suggests that diet termination may have harmful effects (MA- TALON and MICHALS 1991). In addition to the effects of PKU on the affected individual, teratogenic effects result when pregnancy in a woman with PKU is not treated with the PHE-restricted diet. In this situation, called maternal PKU, the developing heterozygous fetus is likely to be severely compromised: a high incidence of microcephaly, mental retardation, con- genital heart defects, and other abnormalities are observed (LEVY and WAISBREN 1983). By contrast, strict regulation of PHE intake from the onset of pregnancy allows normal fetal development (SCRIVER, KAUFMAN and Woo 1989; LENKE and LEVY 1982).

The biochemistry and genetics of PKU are well understood. PKU patients display at least 20-fold el- evated serum PHE and usually <1% activity of PAH [McKusick no. 26160; EC number 1.14.16.1 (KAUF- MAN et al. 1975)]. The genetic locus encoding this enzyme has been mapped, and the PAH coding se-

' Present address: Department of Biological Sciences, The Wichita State University, Wichita, Kansas 67208-1595.

Genetics 134: 1205-1210 (August, 1993)

quence has been cloned in both human (Woo et al. 1983) and mouse (KWOK et al. 1985). Specific DNA alterations have been found for many human PKU mutations (LEDLEY et ul. 1990). Despite these ad- vances, little is known about certain aspects of the biology of PKU. Why is PAH deficiency so damaging to the developing brain? Is PHE the ultimate inhibitor of postnatal central nervous system development and the teratogen in maternal PKU (GUTTLER et al. 1987; LEVY 1988)? Can methods for somatic gene therapy be developed? Clearly, an animal model would be invaluable in studies aimed at addressing these ques- tions.

For some time attempts have been made to produce rat PKU models by administering PHE analogs that inhibit PAH activity (FIGLEWICZ and DRUSE 1980; HUETHER and NEUHOFF 1981; LANE et al. 1980). However, the interpretation of such experiments is complicated by side effects of the inhibitors. Thus, a genetic animal model with the same defect as in the human PKU patient is more appropriate for studying the biology of PKU.

We have reported germline ethylnitrosourea (ENU) mutagenesis followed by a PHE clearance screen to isolate a mouse mutant line, PAHhph-5, deficient in PAH activity (MCDONALD et al. 1990). This first ro- dent mutant fails to exhibit the biological effects as- sociated with human PKU. For clarity and consistency of nomenclature, we are renaming this line PAH'""' (enul). Subsequent mutant alleles, also induced by ENU, will be Pah""", PaVnu3, etc. The present com- munication describes the results of a second round of mutational analysis, with the consequent isolation of two new mutant alleles of Puh. The new lines display

1206 A. Shedlovsky et al.

numerous biological characteristics of human PKU and illustrate one way of perfecting mouse models.

MATERIALS AND METHODS

Mice and diets: Animals were maintained and bred by standard methods of mouse husbandry (LES 1966) using Teklad Breeder Blox containing 20% protein. The PHE deficient diet was Teklad Research Diet #TD90363, con- taining all essential amino acids except PHE; with sucrose (490.88 g/kg); corn starch (150 g/kg); corn oil (100 g/kg); cellulose fiber (30 g/kg); mineral mix, AIN-76 (35 g/kg); calcium phosphate, dibasic (4.5 g/kg); vitamin mix, Teklad 40060 (10 g/kg); and ethoxyquin as an antioxidant (0.02 g/ kg). The BTBR strain has been described (SHEDLOVSKY et al. 1986).

Biochemical determinations: Serum was prepared from blood collected from the retroorbital sinus into heparinized capillary tubes. PHE concentrations were determined by a fluorometric assay, described in Sigma Diagnostics Proce- dure No. 60-F (MCCAMAN and ROBINS 1962). Urinary ke- tones (level of detection 15 mg/dl) were measured by the color produced by a drop of urine on Phenistix Reagent Strips (Miles Inc. Diagnostic Division, Elkhart, Indiana 465 15).

Phenylalanine hydroxylase assays: Crude liver lysates were prepared and assayed as previously described (MC- DONALD et al. 1990). In brief, tissue was homogenized in 150 mM KC1/0.7 mM 2-mercaptoethanol (pH 7.0) and clar- ified by centrifugation at 14,000 X g in the cold for 15 min. PAH activity was measured by the phenylalanine-dependent oxidation of NADH in excess quinonoid dihyropteridine reductase and 6-methyltetrahydrobiopterin. The rate of oxidation was linear for at least 15 min and was directly proportional to total protein concentration over a range of at least 0.6-1.4 mg/ml reaction volume. The activities pre- sented are the total values minus the background rate rt standard deviation. Each value is the average of at least four determinations.

Protein: Concentrations were determined by the Biuret method.

Western blot analysis: Flash frozen liver portions, about 100 mg wet weight, were homogenized on ice in 2 ml RIPA buffer (SUTRAVE, KELLY and HUGHES 1990) containing the proteinase inhibitors aprotinin and phenylmethylsulfonyl fluoride. Samples were electrophoresed at 6 rg/lane in a 15% SDS/polyacrylamide gel (PAGE), with 0.25 pg rat PAH (Sigma) as control, and immunoblotted as described by THOMPSON et al. (1 989).

RNase protection assay: Total liver RNA was isolated (CHOMOZYNSKI and SACCHI 1987) from flash frozen livers. The PAH cDNA was cloned into a pGEM7ZF(+) vector and digested with the restriction endonuclease, DraI. "P-La- beled PAH riboprobe was prepared by transcription from the 3'-end of the cDNA to yield a 358-bp fragment. Assays were carried out at 37" as described by GREENBERG (1987) with only ribonuclease T1. The nuclease-protected frag- ment is expected to be 3 15 bp.

RESULTS

Isolation of PAH mutants: Inbred BTBR males were treated with the germline mutagen ENU and mated to normal BTBR females, as previously de- scribed (LES 1966). Three hundred and fifty progeny of this cross were each screened as follows for muta-

tions that fail to complement Pahcnu'. T h e potential carrier was mated to a PahCnU1 homozygote, and neo- nates were tested for hyperphenylalaninemia by the Guthrie test (GUTHRIE and SUSI 1963). When a pro- portion of the test progeny was found to be hyper- phenylalaninemic, the carrier parent was mated with a normal BTBR partner, and a congenic mutant line was subsequently established. We previously found (unpublished) that neonates homozygous for Pahmur display a very slight Guthrie-positive phenotype. Using the present strategy, we hoped to identify new mutations more severely affected and displaying a PKU phenotype. Among the 350 potential carriers, each representing one mutagenized paternal gamete, we isolated two new alleles, Pahcnu2 and Pahcnu3, each with more pronounced phenotypes than the original.

Noncomplementation does not itself prove allelism; cases have recently been described in which recessive alleles at distinct loci fail to complement (see STEARNS and BOTSTEIN 1988). Indeed, the high frequency of recovery of these mutations [nearly 10 times that observed per locus in other murine genes (RUSSELL et al. 1979)] raises the question of whether they are each alleles of PahC""'. Tight genetic linkage was demon- strated by scoring the phenotype of progeny obtained by intercrossing F1 mice doubly heterozygous for both PahCnul and either PahCnU2 or Pahcnu3. At 3 days of age a drop of tail blood was tested for hyperphenylalani- nemia with the Guthrie assay (GUTHRIE and SUSI 1963). In questionable cases, the mutant phenotype was confirmed at weaning with a PHE clearance test (MCDONALD et al. 1990). Among over 100 progeny from each of these two crosses, all exhibit a mutant phenotype. If these noncomplementing mutations were unlinked to PahCnU1, one would expect a normal phenotype in 5/16 of the progeny. Since none were observed, both of these mutations map within 5.6 cM of Pah""" with 95% confidence; we assume that they are alleles. We have intercrossed the PAHenU2 and PAH'""' lines to produce double heterozygotes. These display the same characteristics as each homozygous mutant line.

Biochemical and whole animal characterization: The enu l neonates and adults were only slightly hy- perphenylalaninemic. By contrast, the enu2 and enu3 adults had a 10-20-fold elevated serum PHE level (Table 1). Urine has been tested for phenylketones with Phenistix paper strips: wild type was negative (<15 mg/dl), enul was slightly elevated, and enu2 and enu3 were more severely elevated (Table 1). Further, in crude liver extracts, PAH activity was markedly deficient in each mutant animal (Table 2). Mixing experiments (not shown) demonstrated that mutant extracts do not contain an inhibitor of PAH activity. Western blot analysis, with a polyclonal anti- PAH antiserum (FRIEDMAN, LLOYD and KAUFMAN

Mouse PKU 1207

TABLE 1

Serum PHE and urinary ketone concentrations in adult wild- type and mutant strains

~~~ ~ ~~

Strain No. of animals [PHE] [KETONE]

On Breeder Blox (20% protein) Wild-type 20 1.5 + 0.5 neg(2) enu 1 50 2.8 1.W neg(l), 15 (2) enu2 100 23.1 -C 6.9 15-40 (2), 100 (I)* enuf 10 20.3 f 5.9 15-40 (3), 100 (3)*

On defined diet supplemented with 100-125 mg/dl PHE in the drinking water

enu2 15 8.0 + 6.6 15-40 (2) enu3 2 7.1 + 4.7 Not tested

PHE concentrations reported are the average, + the standard deviation, of animals maintained on the indicated diet for at least 2 weeks prior to sampling. Values reported are all mg/dl. The number in parentheses indicates the number of animals tested.

This value is significantly different from wild type (P < 0.0005, Wilcoxon rank sum test). * In addition, these have been shown to contain elevated levels of phenyl lactic and phenyl pyruvic acids u. WOLFF, unpublished).

TABLE 2

PAH activity in wild-type and mutant liver extracts

Strain (A O.D. X IO'/min/mg protein) Liver PAH activity

Wild-type enu 1 enu2 enu3

40 + 6 5 * 5 O k 2 5 + 5

Protein: PAH + e3 e3 e2 e l + PAH Lanes: 1 2 3 4 5 6 7 8

P u p . - .

FIGURE 1.-Westem blot analysis of crude liver lysates of wild type and mutants. PAGE analysis for PAH cross-reactive proteins was carried out on standard concentrations (6 pg) crude liver lysates of wild type (+) and the three HPH mutants: enul (el), enu2 (e2), enuf (e3). Lanes 1 and 8 contained 10 pg rat liver PAH (Sigma). Control lanes 1-3 were developed without anti-PAH antibody. Lanes 4-8 were developed after treatments with polyclonal sheep anti-rat PAH antibody (FRIEDMAN, LLOYD and KAUFMAN 1972) followed by alkaline phosphatase conjugated rabbit anti-sheep IgG. Bands were developed by incubating the immunoblot with alkaline phosphatase substrate.

1972), was carried out on crude liver extracts: enu2 contained reduced, but readily detectable, amounts of cross-reacting material (CRM); enu 1 contained lev- els at the borderline of detection, and enu3 levels were below detection (Figure 1).

Analysis of total liver RNA by ribonuclease protec-

RNA: t + t el e2 e3 tRNA Markers pg: 10 1.0 0.1 10 10 10 10

I. -., .. . - -369

1 -31 1

-281 -271

FIGURE 2.-RNase protection assay. The unprotected size of the radioactive fragment is 358 nucleotide bases; after hybridization to PAH message, the size of the RNase-protected fragment is 315 nucleotide bases. RNA from wild type (+) was used at 10, 1.0 and 0.1 pg. All other RNAs were used at 10 pg: enul (el), enu2 (e2). enu3 (e3). and yeast tRNA as a control. The numbers to the right of the figure are the size, in nucleotides, of the nucleic acid marker bands.

tion (see MATERIALS AND METHODS) demonstrated that each mutant produced PAH mRNA. The enul mice produced normal steady state RNA levels, while those in enu3 were reduced to about 10% and in enu2 to less than 1% (Figure 2).

We have compared homozygotes with their heter- ozygous littermates emerging from crosses of carrier female by mutant male. PAH'""* and PAHenU3 animals grew slowly, typically had small heads, and displayed behavioral abnormalities starting at about 2 weeks and lasting into adulthood. While grooming them- selves, young mutants frequently toppled over; in swim tests they were relatively uncoordinated; and, finally the mutants appeared less alert than their het- erozygous siblings. Strikingly, like human PKU pa- tients (JERVIS 1937), the mutant animals displayed pronounced hypopigmentation, beginning at about 2 weeks of age and becoming more severe until stabiliz- ing at about 5 or 6 weeks. Under conditions of ele- vated PHE (e.g., at 25 g/dl in the drinking water), we observed a reduced growth rate of pre- and post- weaning enu 1 pups (unpublished). However, such an- imals function normally, even on long-term expo- sures. By contrast, both enu2 and enu3 animals sicken within days under these conditions.

Maternal effect: In contrast to enul , enu2 and enu3 females exhibited a severe maternal effect under normal husbandry. On our standard breeding diet (see MATERIALS AND METHODS) (about 1% PHE), the average litter size was close to normal, but no pups survived beyond several hours (Table 3). Like mater- nal PKU in humans, this effect depended entirely upon the genotype of the mother. Therefore, we have pooled data from crosses in which the paternal geno- type was varied.

To find a diet permissive for breeding the new mutants as homozygotes, we started with a solid de- fined synthetic diet lacking PHE, which has been added to the drinking water at different concentra- tions. At 125 mg/dl PHE, the serum PHE levels and

1208 A. Shedlovsky et al.

TABLE 3

Maternal effect of enu2 and enu3 ~

Maternal genotype

+/+ Pahnd/+ PahnY2/PahMU2

On standard diet (20% protein) No. of litters 48 1 1 12 No. of progeny at

Birth 365 110 3 2a Weaning 347 98 0

+/+ Pahn*/PahnY2 Pah"'3/Pah-"Y

On defined diet supplemented with 100-125 mg/dl PHE in

No. of litters 7 9 2 No. of progeny at

Birth 39 47 8 Weaning 16 18 3

the drinking water

Matings were set up after females had been maintained on the indicated diet for at least 1 week. At weekly intervals females were palpated to detect pregnancies. Close to the expected end of the gestation period, females were checked daily for litters. The num- ber of progeny were counted within 12 hr of birth and again at weaning.

a Total from six litters observed at birth.

TABLE 4

Effects of maternal genotype and diet on progeny survival

Maternal genotype

(1) (2) (3) Gestation . . . . . . . . PahnY2/Pahmd Pah"'2/Pah"'2 Postnatal . . . . . . . . Pahm"2/Pah"'d +I +

No. of litters 9 9 12 No. of progeny at

Birth 55 40 86 Weaning 54 9'" 80

On defined diet as in Table 3, during gestation and switched to standard diet within 12 hr of birth. Column ( l ) , normal pups with wild type mothers; column (2), pups with enu2 mothers; column (3), pups from enu2 mothers, fostered to wild type within 12 hr of birth.

Progeny from only two litters.

urinary ketones are only slightly elevated (Table 1) and hypopigmentation is partially reversed. This last effect is similar to that observed in human PKU pa- tients on low protein diets (BICKEL, GERRARD and HICKMANS 1954; WOOLF, GRIFFITHS and MONCRIEFF 1955). Under these conditions, the mutant and wild type display similar breeding performances (Table 3). However, in each case, neonates are gradually lost during the first week until an average litter size of about two remains. If the wild-type mother is switched to the standard diet on the day of birth, progeny survival is normal (Table 4). The results obtained with litters from mutant mothers (Table 4) are dramatically different: progeny from only two of nine litters sur- vived. By contrast, if progeny from mutant mothers are fostered at birth to wild-type mothers (on standard diet), survival is essentially normal.

TABLE 5

Effects of maternal genotypes on progeny survival

Maternal genotype

+I+ PahnU2/Pahmu2

No. of litters 6 12 No. of progeny

Fostered 54 92 Weaning 45 1 7"

Timed matings were carried out on standard diet. Pups were taken by Caesarean section on day 19 postcoitum and fostered to normal mothers who had delivered their litters within the previous 12 hr. Mutant females were mated to carrier males and progeny were genotyped at weaning.

" Nine homozygotes and eight heterozygotes.

We wished to test the effect of standard diet on prenatal development in mutant mothers. In addition, we wanted to know if progeny genotype was relevant to any detected dietary effects. Therefore, mutant females were bred to carrier males (on standard diet) and examined daily for vaginal coital plugs. On the 19th day of gestation pregnant females were sacri- ficed, pups were removed and fostered to wild-type mothers on standard diet. Survival, to weaning age, of pups from control, wild-type females, was about 90% (45 of 54), see Table 5 . By contrast, survival from mutant females was about 20% (1 7 of 92) with no effect of progeny genotype.

DISCUSSION

The major result that we report is the establishment of two mutant mouse strains that each closely simulate human PKU. PAH enzyme is much reduced, and blood PHE and urinary ketones are correspondingly increased unless the dietary PHE intake is restricted. We observe evidence for abnormal central nervous system development, and will seek an objective meas- ure of this effect. The mutants exhibit a severe ma- ternal effect on standard breeding diet. We have not determined the cause of neonatal death and, there- fore, we are not yet in a position to evaluate how closely these maternal effects reflect those described in human maternal PKU. However, this lethal effect is avoided when dietary PHE is low during gestation and normal from birth. Further studies are in progress to explore the effects on development of the pre- and postnatal metabolic environment.

The enul strain was selected on the basis of its inability to clear a load of injected PHE. Since we have not observed any other mutant phenotype in vivo, we conclude that the mutation is "leaky." There- fore, it was surprising to find that crude liver extracts of enul animals contained very little PAH enzymatic activity or CRM. Perhaps these animals make an al- tered enzyme that is rapidly degraded -in vitro but more stable in vivo, resulting in sufficient catalytic

Mouse PKU 1209

activity to catabolize the PHE provided in the stand- ard diet.

Overall, we find no correlation between the amount of PAH-CRM and the mutant phenotype in the whole animal. In addition, the amounts of PAH-CRM and mRNA vary independently of each other. We find that enu2, the mutant with the most PAH-CRM, contains the lowest level of mRNA. Perhaps these unexpected results will be better understood when the molecular nature of each mutation within the Pah locus has been determined.

The mutants described here provide material that is necessary to study several aspects of PKU biology. We assume that it was purely fortuitous that the initial PAH mutant isolate was less suitable as a PKU model than the two new alleles. However, this experience underlines considerations that may be crucial in other mouse germline mutagenesis experiments. To obtain a broad spectrum of recessive mutant alleles at a locus, a two-step approach is most effective. The first step, generally requiring a three-generation pedigree screen, is logistically demanding. But the second step, a noncomplementation or “locus-specific” screen (RUSSELL 1951; LYON, PHILLIPS and BAILEY 1972), is more facile. In the absence of a holding generation, a mutagenized gamete is tested by a single testcross animal; when a holding generation is desired, as de- scribed here, a mutagenized gamete can be tested by one litter. Thus, both time and space are dramatically reduced compared with the three-generation scheme. These considerations are particularly important for genes with vital function. By necessity, the initial recessive allele must be “leaky” to permit viability. The second stage then identifies mutant alleles that may include recessive lethals. By contrast, the vital gene can first be mutated by gene targeting (for example, see TYBULEWICZ et al. 1992) and the null mutant allele propagated in heterozygous form. Next, mutagen-induced noncomplementing viable alleles can be readily isolated by a specific locus screen. Such a strategy may yield mouse models that accurately reflect the human disease phenotype (WILSON and COLLINS 1992; DAVIES 1992).

Because of their strong phenotype, these mutants can be directly utilized in exploring methods of so- matic gene therapy. Since the mutations are congenic on an inbred strain of mice, one could directly intro- duce wild-type hepatocytes into mutants and assess correction of the mutant phenotype in the whole animal. Alternatively, one could employ virally me- diated gene therapy by infecting mutants with a retro- virus containing Pah cDNA. Since many inborn errors of metabolism result from hepatic enzyme deficien- cies, we expect that methods developed with these mutants will have implications even broader than those applying to PKU.

We thank M. CORONAW for excellent technical assistance with the Western blot analysis, J. WOLFF for urinary organic acid analysis, M. WOCH for care in breeding and maintaining mice, and J. F. CROW, C. DENNISTON, 0. N E L ~ ~ N and I. RIEGEL for critical reading of the manuscript. This work was supported by the National Insti- tute of Diabetes and Digestive and Kidney Diseases (DK40393); training grants GM07133 from the National Institute of General Medical Sciences and CA09230 from the National Cancer Institute; a graduate fellowship to D.S. from the National Science Foundation and National Cancer Institute Core Grant CA07175 to the McArdle Laboratory. This is publication no. 3327 from the Laboratory of Genetics.

LITERATURE CITED

BICKEL, H., J. GERRARD and E. M. HICKMANS, 1954 Influence of phenylalanine intake on the chemistry and behaviours of a phenylketonuric child. Acta Paediatr. 43: 64-73.

CHOMOZYNSKI, P., and N. SACCHI, 1987 Single-step method of RNA isolation by acid guanidinium thiocyanate-phenolchlo- roform extraction. Anal. Biochem. 162 156-159.

DAVIES, K., 1992 Mulling over mouse models. Nature 3 5 9 86. FIGLEWICZ, D. A., and M. J. DRUSE, 1980 Experimental hyper-

phenylalaninemia: effect on central nervous system myelin subfraction. Exp. Neurol. 67: 315-329.

FRIEDMAN, P. A., T. LLOYD and S. KAUFMAN, 1972 Production of antibodies to rat liver phenylalanine hydroxylase: cross re- activity with other pterin-dependent hydroxylases. Mol. Phar- macol. 8: 501-510.

GREENBERG, M. E., 1987 Ribonuclease Protection Assay, pp. 4.7.1-4.7.8 in Current Protocols an Molecular Biology I , edited by F. M. AUSUBEL, R. BRENT, R. E. KINGSTON, D. D. MOORE, J. G. SEIDMAN, J. A. SMITH and K. STRUHL. Greene Publishing Associates and Wiley Interscience, New York.

GUTHRIE, R., and A. SUSI, 1963 A simple phenylalanine method for detecting phenylketonuria in large populations of newborn infants. Pediatrics 32: 328-343.

GUTTLER, F., A. G. DILELLA, F. D. LEDLEY, A. S. LIDSKY, S. C. M. KVOK and S. L. WOO, 1987 Molecular biology of phenylke- tonuria. Eur. J. Pediatr. 146: A5-All.

HUETHER, G., and V. NEUHOFF, 1981 Use of alpha-methylphen- ylalanine for studies of brain development in experimental phenylketonuria. J. Inherited Metab. Dis. 4 67-68.

JERVIS, G. A., 1937 Phenylpyruvic oligophrenia. Arch. Neurol. Psychiatry 38: 944-963.

KAUFMAN, S., N. A. HOLTZMAN, S. MILSTIEN, I. J. BUTLER and A. KRUMHOLZ, 1975 PKU due to a deficiency of dihydropteri- dine reductase. N. Engl. J. Med. 293: 785-790.

KWOK, S. C. M., F. D. LEDLEY, A. G. DILELLA, K. J. H. ROBSON and S. L. C. WOO, 1985 Nucleotide sequence of a full-length cDNA clone of human phenylalanine hydroxylase. Biochemis- try 2 4 556-56 1.

LANE, J. D., B. SCHOENE, U. LANGENBECK and V. NEUHOFF, 1980 Characterization of experimental phenylketonuria: aug- mentation of hyperphenylalaninemia with alpha-methyl phen- ylalanine and p-chlorophenylalanine. Biochim. Biophys. Acta

LEDLEY, F. D., H. E. GRENEIT, B. S. DUNBAR and S. L. C. WOO, 1990 Mouse phenylalanine hydroxylase: homology and diver- gence from human phenylalanine hydroxylase. Biochem. J.

LENKE, R. R., and H. L. LEVY, 1982 Maternal phenylketonuria: results of dietary therapy. Am. J. Obstet. Gynecol. 142 548- 553.

LES, E. P., 1966 Husbandry, pp. 29-37 in Biology of the Laboratoy Mouse, edited by E. L. GREEN. McGraw-Hill, New York.

LEVY, H. L., 1988 Maternal phenylketonuria, pp. 227-242 in Transplacental Effects on Fetal Health. Alan R. Liss, New York.

627: 144-156.

267: 399-406.

1210 A. Shedlovsky et al.

LEVY, H. L., and S. E. WAISBREN, 1983 Effects of untreated maternal phenylketonuria and hyperphenylalaninemia on the fetus. N. Engl. J. Med. 309: 1269-1274.

LYON, M. R., R. J. PHILLIPS and H. J. BAILEY, 1972 Mutagenic effects of repeated small radiation doses to mouse spermato- gonia. I. Specific-locus mutation rates. Mutat. Res. 15: 185- 190.

MATALON, R., and K. MICHAIS, 1991 Phenylketonuria: screening, treatment and maternal PKU. Clin. Biochem. 24: 337-342.

MCCAMAN, M. W., and E. J. ROBINS, 1962 Fluorimetric method for the determination of phenylalanine in serum. Lab. Clin. Med. 59: 885-890.

MCDONALD, J. D., V. C. BODE, W. F. DOVE and A. SHEDLOVSKY, 1990 Pahhph-5: a mouse mutant deficient in phenylalanine hydroxylase. Proc. Natl. Acad. Sci. USA 87: 1965-1967.

RUSSELL, W. L., 1951 X-ray-induced mutations in mice. Cold Spring Harbor Symp. Quant. Biol. 1 6 327-336.

RUSSELL, W., E. KELLY, P. HUNSICKER, J. BANGHAM, S. MADDUX and E. PHIPPS, 1979 Specific-locus test shows ethylnitrosou- rea to be the most potent mutagen in the mouse. Proc. Natl. Acad. Sci. USA 76: 5818-5819.

SCRIVER, C. R., and C. L. CLOW, 1980 Phenylketonuria and other phenylalanine hydroxylase mutants in man. Annu. Rev. Genet.

SCRIVER, C. P., S. KAUFMAN and S. L. C. Woo, 1989 The hyper- phenylaninemias, pp. 495-546 in The Metabolic Basisfor Inher- ited Disease, edited by C. R. SCRIVER, A. BEAUDET, W. SLY and D. VALLE. McGraw Hill, New York.

SHEDLOVSKY, A., J. L. GUENET, L. L. JOHNSON and W. F. DOVE, 1986 Induction of recessive lethal mutations in the T/t-H-2

1 4 179-202.

region of the mouse genome by a point mutagen. Genet. Res. 47: 135-142.

STEARNS, T., and D. BOTSTEIN, 1988 Unlinked noncomplemen- tation: isolation of new conditional-lethal mutations in each of the tubulin genes of Saccharomyces cerevisiae. Genetics 119

SUTRAVE, P., A. M. KELLY and S. H. HUGHES, 1990 ski can cause selective growth of skeletal muscle in transgenic mice. Genes Dev. 4 1462-1472.

THOMPSON, N. E., T. STEINBERG, D. B. ARONSON and R. R. BUR- GESS, 1989 Inhibition of in vivo and in vitro transcription by monoclonal antibodies prepared against wheat germ RNA po- lymerase I1 that react with the hepapeptide repeat of eucaryotic RNA polymerase 11. J. Biol. Chem. 264: 1151 1-1 1520.

TYBULEWICZ, V. L. J., M. L. TREMBLAY, M. E. LAMARCA, R. WILLEMSEN, B. K. STUBBLEFIELD, S. WINFIELD, B. ZABLOCKA, E. SIDRANSKY, R. M. MARTIN, S. P. HUANG, K. A. MINTZER, H. WESTPHAL, R. C. M U L L I G A N ~ ~ ~ E. I. GINNS, 1992 Animal model of Gaucher's disease from targeted disruption of the mouse glucoscerebrosidase gene. Nature 357: 407-4 10.

WILSON, J. M., and F. S. COLLINS, 1992 More from the modellers. Nature 359 195-196.

Woo, S. L., A. S. LIDSKY, F. G~JTTLER, T. CHANDRA and K. J. H. ROBSON, 1983 Cloned human phenylalanine hydroxylase gene allows prenatal diagnosis and carrier detection of classical phenylketonuria. Nature 306 151-155.

WOOLF, L. I., R. GRIFFITHS and E. M. MONCRIEFF, 1955 Treatment of phenylketonuria with a diet low in phen- ylalanine. Br. Med. J. 1: 57.

249-260.

Communicating editor: R. E. GANSCHOW