Dwarfism and low insulin-like growth factor-1 due to dopamine depletion in Pts-/- mice rescued by feeding neurotransmitter precursors and H4-biopterin

The tetrahydrobiopterin (BH4) cofactor is essential for the biosynthesis of catecholamines and serotonin and for nitric-oxide synthase (NOS). Alterations in BH4 metabolism are observed in various neurological and psychiatric diseases, and mutations in one of the human metabolic genes causes hyperphenylalaninemia and/or monoamine neurotransmitter deficiency. We report on a knockout mouse for the Pts gene, which codes for a BH4-biosynthetic enzyme. Homozygous Pts-/- mice developed with normal morphology but died after birth. Upon daily oral administration of BH4 and neurotransmitter precursors the Pts-/- mice eventually survived. However, at sexual maturity (6 weeks) the mice had only one-third of the normal body weight and were sexually immature. Biochemical analysis revealed no hyperphenylalaninemia, normal brain NOS activity, and almost normal serotonin levels, but brain dopamine was 3% of normal. Low dopamine leads to impaired food consumption as reflected by the severe growth deficiency and a 7-fold reduced serum insulin-like growth factor-1 (IGF-1). This is the first link shown between 6-pyruvoyltetrahydropterin synthase- or BH4-biosynthetic activity and IGF-1. Abstract The tetrahydrobiopterin (BH4) cofactor is biosynthesis of catecholamines and serotonin, and for NOS. Alterations in BH4 metabolism are observed in various neurologic and psychiatric diseases, and mutations in one of the human metabolic genes neurotransmitter deficiency. We report on a knockout mouse for the Pts gene, which codes for a BH4-biosynthetic enzyme. Homozygous Pts -/- mice developed with normal morphology but died after birth. Upon daily oral administration of BH4 and neurotransmitter precursors the Pts -/- mice eventually survived. However, at sexual maturity (6 weeks) the mice had only one third of the normal body weight and were sexually immature. Biochemical analysis revealed no hyperphenylalaninemia, normal brain NOS activity, and almost normal serotonin levels, but brain dopamine was 3% of normal. Low dopamine leads to impaired food consumption as reflected by the severe growth deficiency and a 7-fold reduced serum insulin-like growth factor-1 (IGF-1). This is the first link shown between PTPS- or BH4-biosynthetic activity and IGF-1.


INTRODUCTION
Tetrahydrobiopterin (BH 4 ) plays a central essential role in metabolism, involving monoamine neurotransmitter biosynthesis, hepatic phenylalanine degradation, and nitric oxide (NO) production. The absolute requirement of BH 4 for such enzymatic functions is reflected by severe disturbances or even lethality in the case of cofactor limitation due to mutations in BH 4 -metabolic genes. Patients with cofactor deficiency may exhibit severe dopamine and serotonin neurotransmitter deficiency, hyperphenylalaninemia, and reduced NO production (1,2). The BH 4 cofactor is synthesized from guanosine triphosphate (GTP) by a set of reactions involving three enzymes. GTP cyclohydrolase I (GTPCH), the first enzyme in BH 4 biosynthesis, catalyzes the formation of dihydroneopterin triphosphate from GTP (3). GTPCH activity is regulated by its substrate GTP, BH 4 , and phenylalanine. A physiological mechanism for posttranslational control of GTPCH activity involves feedback inhibition by BH 4 . Notably, feedback inhibition results from BH 4 -induced complex formation of GTPCH with a regulatory protein known as GTPCH feedback regulatory protein (GFRP) (4)(5)(6). In the second step, the 6-pyruvoyl-tetrahydropterin synthase (PTPS) catalyzes the conversion of dihydroneopterin triphosphate to 6pyruvoyl-tetrahydropterin. PTPS needs to be phosphorylated to be fully active (7,8). Sepiapterin reductase (SR) is required for the final step reductions of the diketo intermediate, 6-pyruvoyl-tetrahydropterin to BH 4 . BH 4 is required as cofactor for phenylalanine hydroxylase (PAH), tyrosine hydroxylase, and tryptophan hydroxylase. The latter two are key enzymes in the biosynthesis of the neurotransmitters dopamine and serotonin (9). The complete hydroxylating system of aromatic amino acids consists of the two additional BH 4 -regenerating enzymes: pterin 4-α-carbinolamine dehydratase (PCD) and dihydropteridine reductase (DHPR) (10). BH 4 is also required for the nitric oxide synthase enzymes (11).
A deficiency of phenylalanine catabolism, leading to hyperphenylalaninemia (HPA), comprises a heterogeneous group of disorders caused by a partial or complete deficiency of the hepatic apoenzyme PAH, or by one of the enzymes involved in cofactor biosynthesis (GTPCH or PTPS) (12,13), or regeneration (DHPR and PCD) (14)(15)(16). Whereas severe HPA, leading to classical phenylketonuria, can only be treated with phenylalanine-low diet, patients with BH 4 -responsive PAH deficiency can be treated with BH 4 alone (17). Two disorders of BH 4 metabolism may present without HPA. These are dopa-responsive dystonia (DRD or Segawa disease) (18) and sepiapterin reductase deficiency (19,20). While DRD is caused by a mutation in the GTPCH gene and is inherited in an autosomal dominant manner, SR deficiency is an autosomal recessive trait. Both diseases manifest severe biogenic amine deficiencies.
To diagnose BH 4 deficiencies and follow-up of the resulting pathologies, only limited possibilities are available, including measurements of metabolites in body fluids, and follow-up of the disease development almost exclusively under treated conditions. Diagnosis starts in most cases with screening of HPA with the Guthrie card and determination of plasma phenylalanine levels as an indirect measurement of hepatic phenylalanine hydroxylase (PAH) activity. Analysis of phenylalanine and tyrosine in serum or plasma before and after a BH 4 challenge are often applied as an additional diagnostic tool for differentiation between classical phenylketonuria and biopterin variants. Furthermore, urinary pterin analysis and enzymatic measurements in erythrocytes or skin fibroblasts are carried out to gather information on biopterinmetabolizing enzymes. These data are then combined with a neurotransmitter status. For neurotransmitters, the dopamine and serotonin neurotransmitter degradation products homovanillic acid (HVA) and 5-hydroxyindoleacetic acid (5-HIAA), respectively, are measured in cerebrospinal fluid (CSF), thus following the activities of tyrosine and tryptophane hydroxylases. Furthermore, NO metabolites at least for brain NO-synthase (NOS) isoenzyme activity are determined in CSF. Symptoms of BH 4 deficiency include a vast range of abnormalities of the central nervous system, including microcephaly, seizures, hypertonia, hypersalivation, temperature instability, feeding difficulties, and mental retardation. The goals of treatment are to control HPA by dietary restriction of phenylalanine (in PAH deficiency) or BH 4 administration (in GTPCH and PTPS deficiency), and to restore neurotransmitter homeostasis by oral administration of the dopamine and serotonin precursors L-Dopa and 5-hydroxytryptophan, respectively, in BH 4 deficiencies. Late detection and introduction of treatment leads to irreversible brain damage. For patients with BH 4 deficiency, HPA is controllable with oral doses of 2-10 mg synthetic BH 4 /kg/day. However, such relatively low doses of BH 4 do not allow the cofactor to penetrate the blood-brain barrier efficiently (21,22). To some extent, this problem can be overcome by administering higher doses of BH 4 , up to 20 mg/kg/day, together with corresponding neurotransmitter precursors (23). The combined therapy is mandatory to avoid neurological damage; however, this treatment is not sufficient in every case (24). In order to analyze in more detail the consequences of BH 4 deficiency and its treatment, and to study pathologies in the organism, the use of animal models is required. Here we report on the generation of an animal model for PTPS deficiency by knocking out the Pts gene in the mouse. This led to perinatal lethality of otherwise normal born animals. Treatment studies with daily oral administration of different concentrations of BH 4 , L-Dopa, and 5-hydroxytryptophan for BH 4 led to the observation, that mice can be rescued but exhibit severe growth deficiency leading to dwarfism due to low serum insulin-like growth factor-1 (IGF-1).

EXPERIMENTAL PROCEDURES
Pts gene targeting A genomic clone containing the Pts gene encoding the mouse PTPS was isolated from a 129/Sv-λ phage library and characterized previously (25). To construct a targeting vector, a KpnI-NcoI fragment, generated by PCR and spanning exon 1 and the first 9 codons from exon 2 were used for the short arm of homology (Fig. 1A). Exon 2 of this fragment was ligated in-frame with an NcoI-BamHI fragment containing the prokaryotic lacZ gene, followed by a phosphoglycerate kinase promoter (Pgk)-neo cassette. A Pgk-tk cassette was added 5' to this short arm of homology. The long arm of homology was a 5.4 kb HindIII fragment containing exons 5 and 6 of the Pts gene. The final targeting vector, plasmid pMSY23, was linearized, electroporated into 129/Sv embryonic stem (ES) cells and selected for G418 and FIAU resistance as described (26).
For PCR screening of ES clones, a nested PCR with two rounds of 40 cycles under standard amplification conditions with an annealing temperature of 55°C was applied.
For the first PCR, the 5' primer MSY69 annealed outside (upstream) of the short arm of homology, and the 3' primer PLACZ6 matched to the lacZ gene. For the second round of amplification, the 5' primer MSY70 was upstream of exon 1, and the 3' primer  Figure 1B). β-Galactosidase activity was determined with extracts from ES cells according to a published protocol (27).

Replacement therapy
For an overview of concentrations of drugs used for treatments, see the dissolved compounds were analyzed by standard HPLC (see below). Aliquots were kept frozen, thawed before use, and mixed for the amount required according to Table   1. The freshly prepared solutions were orally administered using yellow tips and a Gilson micropipette. For the 'medium' and 'high' treatment protocols, the daily aliquots were divided into 2 daily doses.

Neopterin and biopterin measurements
A volume of 50 µl of liver or 50 µl of brain tissue homogenates were adjusted to 100 µl, and oxidized with 10 µl of oxidizing solution (5 g/l iodine and 10 g/l potassium iodide in 1 M HCl). After oxidation in the dark for 60 minutes, the reaction was stopped by adding 10 µl of freshly prepared ascorbic acid (20 g/l). A total of 14 µl of 1 M NaOH was added to adjust the mixture to pH 8.5, followed by incubating with 20 µl of an alkaline phosphatase solution at 37°C for 1 hour (300 U/ml calf intestine alkaline phosphatase from Roche in 0.1 M Tris-HCl, pH 8.0, 1 mM MgCl 2 , 0.1 mM ZnCl 2 ).
The mixture was adjusted to pH 2 by adding 5 µl of 2 M HCl, and deproteinized through an Ultrafree-MC filter (Millipore). Neopterin and biopterin are measured from the filtrate by HPLC (28). The concentrations are expressed as pmol per mg of protein.

Enzymatic assays
A volume of 100 µl tissue homogenates were desalted on a spin column (MicroSpinTM G-25 columns, Amersham Pharmacia Biotech), and 100 µg of protein from the liver filtrate or 200 µg of protein from the brain filtrate were used for GTPCH and PTPS assays, respectively. GTPCH assay: a final volume of filtrate was adjusted to 50 µl and was added to 148 µl homogenizing buffer and 2 µl of 100 mM GTP (Roche). This mixture was divided into two 100-µl portions. One portion was immediately oxidized as blank with cell extract and the second portion was incubated for 60 minutes at 37°C. The reaction was stopped by cooling the sample on ice and adding 10 µl of oxidizing solution (5 g/l iodine and 10 g/l potassium iodide in 1 M HCl). After oxidation in the dark for 60 minutes, the reaction was stopped by adding 10 µl of 20 g/L ascorbic acid (freshly prepared). The mixture war adjusted to pH 8.5 by adding 14 µl of 1 M NaOH, and the sample was incubated with 20 µl of alkaline phosphatase solution at 37°C for 1 hour (300 U/ml calf intestine alkaline phosphatase, Roche; see above). The mixture was adjusted to pH 2 by adding 5 µl of 2 M HCl, and deproteinized through an Ultrafree-MC filter (Millipore). Neopterin was measured from the filtrate by HPLC. One unit of GTPCH produces 1 µmol of neopterin per minute at 37°C. PTPS assay: a final volume of filtrate was adjusted to 50 µl and was added to 60 µl of reaction mixture (100 mM Tris-HCl, pH 7.4, 10 mM MgCl2, 1 mM NADPH, 1 mM NADH, 3 mU of SR, 140 mU of DHPR from Roche, and 60 µmol/L dihydroneopterin triphosphate) in a final volume of 110 µl. This mixture was divided into two portions, one 50-µl aliquot was incubated for 2 hours at 37°C and another 50 µl was used as a blank. A blank without cell extract was incubated at the same time; it contained 50 µl of reaction buffer and 60 µl of reaction mixture. The reaction was stopped by adding 15 µl of 300 g/l trichloroacetic acid (TCA) for protein precipitation, and cooling on ice for at least 10 minutes, followed by oxidation with 5 mg MnO (manganese oxide) for 15 minutes in the dark. For the blanks, the same procedure was used. After 2 minutes of centrifugation at 15'000 g, the supernatant was deproteinized through an Ultrafree-MC filter (Millipore) and analyzed by HPLC. One unit of PTPS produces 1 µmol of biopterin per minute at 37°C. NO assay: the NO, which is the product of NOS, is extremely reactive and undergoes a series of reaction. Nitrite (NO 2 -) and Nitrate (NO 3 -) are the final products. The sum of these two products (nitrite+nitrate) was measured using a commercial Colorimetric Assay Kit (Cayman Chemical, Ann Arbor, MI, USA). During this assay, Nitrate was converted to nitrite utilizing nitrate reductase and measured with the Griess reagent.
Absorbance was read at 570/620 nm in a MicroELISA autoreader MR 530 (Dynatech, Chantilly, VA, USA) Phenylalanine hydroxylase (PAH) assay: the assay was adapted from Ledley et al. (29). Liver homogenate containing 50-100 µg of total protein was used for PAH assay. For the blank, the appropriate amount of liver homogenate was adjusted to a final volume of 104 µl with water, and incubated for 5 min in a 96°C heating block. For the sample, the appropriate amount of liver homogenate was adjusted to a final volume of 77.5 µl with water. A volume of 22 µl of master mixture was added to each sample. The master mixture contained 0.6 mM phenylalanine, 3.6 U catalase (Sigma), 0.15 M KCl in a 0.2 M potassium phosphate buffer, pH 6.8. After pre-incubation at room temperature, the reaction was started by adding 2 µl of 4.5 mM 6-methyl-tetrahydropterin (6M-PH 4 , Shircks Laboratories, Jona, Switzerland) to the samples, and incubated for 60 minutes at 25°C. The reaction was stopped by incubation for 5 minutes in a 96°C heating block, and centrifuged for 5 min at 13000 rpm. The supernatant was filtrated in an Ultrafree-MC filter device and centrifuged again at 5000 g for 15 min. Phenylalanine and tyrosine were quantified with a standard amino acid analyzer (Biochrom 20 Plus, Amersham Pharmacia Biotech).

Phenylalanine concentration in the blood
The blood from the mice was collected on filter paper cards (Guthrie card). Phenylalalnine (and tyrosine) concentrations were measured using electrospray ionization tandem mass spectrometry (ESI-MS/MS).

Protein measurement
Protein concentrations in homogenized tissues, wes determined by the spectrophotometric method described by Bradford, using γ-globulin as a calibrator (30). The activities of the various enzymes are expressed as units per mg protein.

Immunoassays
Blood was collected from 35 and 44-day-old mice at the time animals were sacrificed.
Insulin-like growth factor-1 (IGF-1): serum IGF-1 was separated from clotted blood by centrifugation. It was measured after extraction with acid-ethanol (40 µl serum and 160 µl acid-ethanol). The mixture was incubated for 30 min at room temperature, centrifuged and 100 µl supernatant was diluted 1:6 before analysis. Serum IGF-1 was determined by radioimmunoassay (RIA) using a rat IGF-1 RIA Kit (DSL-2900, Bühlmann laboratories AG, Switzerland). Growth hormone (GH): serum GH was measured by RIA using a specific rabbit antirat antiserum and rat GH as standard. The rat GH RIA Kit (AH R012) was obtained from Bühlmann laboratories AG (Switzerland).
Blood thyroxin (T4): T4 was measured by fluoroimmunoassay using the mouse antithyroxine IgG as first antibody and the anti-mouse IgG as second antibody. The blood was dried on filter paper cards (Guthrie card). The total amount of T4 was determined in the test. The AutoDELFIA TM Neonatal Thyroxine (T4) kit was obtained from Perkin Elmer, Life Sciences, Wallac-ADL AG (Switzerland).

Targeted deletion of the mouse Pts leads to perinatal death
A targeting construct was generated based on the previously isolated and characterized mouse gene structure Pts, encoding the 6-pyruvoyl-tetrahydropterin synthase (25). As shown in Fig. 1A, the pMSY32-targeting vector contained an in-frame lacZ gene fusion at exon 2 of the Pts gene, expressing a putative PTPS-β-galactosidase fusion with 35 N-terminal amino acids from PTPS. Downstream of the LacZ gene, a Pgk-neo cassette was inserted in the opposite direction. Upon correct homologous recombination in ES cells, a putative mutant allele is generated with the lacZ and Pgk-neo inserted, and a deletion of exons 3 and 4, plus most of exon 2. The targeting frequency for correct double cross-over in the 129/Sv ES cells, as verified by PCR, was approx. 5% (not shown). These ES cells had a β-galactosidase activity of 0.05-0.12 OD/mg (wild-type activity <0.001 OD/mg) and a PTPS activity indistinguishable from wild-type (5.4-6.5 µU/mg).
Upon PCR and Southern blot analyses ( Fig. 1B and C), a few Pts +/-ES cell clones were used for blastocyste injection and subsequent generation of PTPS-null mice. Homozygous mice developed with normal morphology in utero, and were born at the expected Mendelian ratio (25% wild-type, 48% heterozygotes, 27% Pts-null mice; n = 159). However, most of the Pts -/mice died within the first hours after birth; at maximum we found 4 mice out of 26 surviving for 7 days. A more detailed analysis of brain development and fine-structure of Pts -/mice will be published elsewhere (in preparation). Southern blot analyses with genomic mouse tail DNA and a 5'-external probe (Fig. 1C), or an internal neo-probe ('Probe B' in Fig. 1A; results not shown) revealed correct homologous recombination at the single mouse Pts gene locus. As compiled in Table 2, newborn knockout mice at day 1 had no PTPS (<0.05 µU/mg) and normal GTPCH activity (0.1-0.3 µU/mg). Heterozygous animals showed intermediate PTPS activity (1.5 µU/mg) compared to normal activity in wild-type mice (8.0 µU/mg). Liver neopterin was almost 200-fold higher than normal (59.0 pmol/mg in Pts -/-), and biopterin was only 4% of wild-type (0.9 pmol/mg). Furthermore, the Pts -/animals had hyperphenylalaninemia with blood values of 1352 µmol/l phenylalanine (normal control levels were between 34-85 µmol/l), and no detectable or very low brain dopamine and serotonin levels. Expression of β-galactosidase was observed in heterozygous and homozygous Pts mutants (not shown). A more detailed study on developmental expression of the PTPS-LacZ fusion will be published elsewhere.

Treatment of Pts -/mice resulted in rescue from lethality but severe dwarfism
In a next step, we thought to rescue the knockout mice by applying a replacement therapy protocol based on the recommended standard concentrations for treatment of human PTPS patients (1). This included daily oral administration of BH 4 to control blood phenylalanine, and the neurotransmitter precursors L-Dopa and 5hydroxyptryptophan. Unexpectedly we learned that with this standard treatment protocol, Pts -/mice survived not longer than for about 3 weeks. We thus extended our treatment studies with three types of application levels (Table 1; see also Experimental Procedures): standard or 'low' treatment, a 'medium' treatment with 2-5-times higher concentrations of BH 4 and neurotransmitter precursors, and a 'high' treatment with roughly 3-10-times higher concentrations. Each treatment group contained 7-14 control animals, i.e. wild-type or heterozygotes, and 3-6 Pts -/mice. The body weight of every single animal was monitored daily, and is depicted for each treatment group in Fig. 2.
As mentioned before, the Pts -/animals with the 'low' treatment did not survive for more than 3 weeks. A similar situation was encountered with the 'medium' treatment, where survival of the Pts -/animals was prolonged but they eventually died between day 31 and 40 after birth (at a certain point knockout animals had to be sacrificed due to progressively poor conditions, and in agreement with the Rules and Guidelines for the Care and Use of Laboratory Animals of the State of Zurich). In contrast, all Pts -/animals with the 'high' treatment survived and were in relatively good health conditions. The experiment was stopped after 6 weeks of treatment, where all animals were sacrificed for biochemical analysis (see below). As shown in Fig. 2A-C, newborn mice regardless of the treatment mode gained weight without significant differences between genotypes for about 2 weeks, and underwent pronounced growth retardation during week 3 (days [15][16][17][18][19][20][21]. Only the Pts -/mice with 'low' treatment seemed to be slightly different from their normal littermates, as they had reduced growth rate almost from birth (see 'low' at day 3 in Fig. 3). This developmental difference in the 'low' treatment group was even more pronounced later, as illustrated also in Figure 2D, which shows a 7-day-old Pts -/animal in comparison with an age-matched heterozygous littermate. After the period of growth stagnation, the Pts -/animals stopped gaining weight independently of the treatment level, whereas all wild-type and heterozygous mice grew normally. The diminutive body size of Pts -/mice was best visible for those that survived due to 'high' treatment, where the body weight was 34% of control at the age of 7 weeks (7.8+1.5 grams for Pts -/mice compared to 23.1+2.4 grams for combined controls; Fig. 3  to light pigmentation, hair loss, and hypothermia. In summary, treatment of Pts -/mice with BH 4 and neurotransmitter precursors resulted in rescue from lethality but severe dwarfism.

Biochemical analysis of sacrificed mice following different treatment protocols
For enzymatic and metabolite analyses in liver, blood, and brain, all groups of treated mice were sacrificed at day 23 for 'low', day 31 for 'medium', and day 44 for 'high' treatment. Liver analysis was also carried out with untreated mice at day 1 after birth (see above), whereas the limited brain material from newborns allowed us to measure only nitrite+nitrate (compare to Tables 2 and 3). As expected, PTPS activity in liver and brain was completely abolished in the Ptsknockout mice, and reduced to roughly 45% in heterozygous animals compared to wild-type. Furthermore, in liver and brain of phenotypically normal mice, i.e. wildtype and heterozygotes, PTPS activity increased with age. Independent of the treatment level, blood phenylalanine, and liver BH 4 and GTPCH activity decreased in 23, 31, and 44-day-old knockout mice compared to controls. Low GTPCH activity in Pts -/mice was surprising since it was expected that hyperphenylalaninemia and low BH 4 -levels, as observed in these mice, results in an increase of GTPCH activity via the stimulatory action of GFRP (6) (see also Discussion).
Brain serotonin was severely lowered only in 23-day-old knockout mice with the 'low' treatment, but approximately 50% of normal in knockouts with 'medium' or 'high' treatment. Instead, brain dopamine was not detectable in knockouts with the 'low' treatment and also very low, between 3-11% of control, in the 'medium' and 'high' treatment. The brain metabolites for dopamine and serotonin, HVA and 5-HIAA, were only slightly reduced in Pts -/mice even under the 'high' treatment conditions, with 52% of normal for HVA and 60% of normal for 5-HIAA (day 44 of treatment). Brain dihydroxyphenlyacetic acid (DOPAC), which is the first degradation product of dopamine following the action of dopamine hydroxylase (or MAO), was indistinguishable among knockout and wild-type or heterozygous animals from the 'medium' and 'high' treatment groups (between 30.2-67.9 pmol/mg; not shown). The NOS activity in brain, as determined by measuring the sum of nitrate+nitrite, revealed no difference among knockout and control animals, and was also independent of treatment levels. NOS activity in the brain of untreated Pts -/or normal mice at day 1 was several-fold higher compared to older animals, but also indistinguishable between the phenotypes (100.0-117.3 nmol/g tissue at day 1). The Pts -/knockout animals, independent of 'medium' or 'high' treatment conditions, exhibited extremely low dopamine and sub-optimal levels of BH 4 and serotonin.
Furthermore, the only metabolic difference we observed among these two treatment conditions, which eventually lead to death of animals with only the 'medium' treatment, was the intermediate HPA.

IGF-1 is severely lowered in rescued Pts -/mice with dwarfism
As the phenotypic characteristics of the dwarf mice might be a consequence of abnormal feeding behavior due to low dopamine and/or of hormonal deregulation, we wondered whether the pituitary growth hormone (GH), the thyroid hormone thyroxin (T4), and the insulin-like growth factor-1 (IGF-1) in serum of knockout animals were reduced. As shown in Fig. 4, the serum IGF-1 levels in the knockout mice were reduced by a factor of 7 in comparison with age-matched controls (knockouts 79+36 ng/ml; controls 541+155 ng/ml). The expected sexual dimorphism between females (482+170 ng/ml) and males (613+113 ng/ml) is also clearly visible as published before (31). Furthermore, we also tested whether the pituitary-derived growth hormone (GH) and the thyroxin-stimulating hormone (TSH)-dependent thyroxin (T4) were also reduced in these animals. However, we found no change in GH and T4 (not shown), indicating that the pituitary gland is normally developed, and thus not the primary reason for dwarfism in these treated Pts -/mice. Low IGF-1 might thus be the biochemical reason for the dwarfism, probably caused by the limiting brain dopamine, as abnormal feeding behavior was reported in dopamine-deficient mice (see Discussion) (32

DISCUSSION
Here we report on treatment studies with a BH 4 -deficient mouse that was generated by targeted disruption of the Pts locus that encodes the second enzyme in the BH 4biosynthetic pathway. This model was sought to study the role of the cofactor in metabolism and treatment. The importance of BH 4 for dopamine and serotonin production has been well established in patient studies, where treatment of cofactor deficiency by replacement with the precursors L-Dopa and 5-hydroxytryptophan is required for neurotransmitter homeostasis and essential for survival (1). The observation that a complete knockout of BH 4 biosynthesis in the mouse leads to a phenotype with perinatal death fits the expectations regarding the absolute requirement of a cofactor with central metabolic importance. The lack of biosynthetic activity for catecholamines, which includes dopamine and norepinephrine, and for serotonin must be one of the primary reasons for the perinatal death, as these neurotransmitters are essential for postnatal survival (33). However, in contrast to humans patients, we were surprised to find that the mice died almost immediately after birth with no visible abnormalities, an observation that was also made by Sumi-Ichinose and co-workers, and published during the course of our study (34).
We found only a few knockout mice surviving for up to seven days after birth, probably due to the BH 4 present in mother's milk (35), while the amount of milk available is in turn dependent on the litter size and/or the mother's behavior. Data from approximately 250 PTPS patients, as compiled in the database www.bh4.org, does not reveal perinatal lethality, and although symptoms may be noted during the neonatal period, abnormalities develop typically during the first weeks of life (1). Furthermore, low birth weight and microcephaly, which is typical for PTPS deficiency in human newborns (36), was not observed in our mice. On the other hand, symptoms like hypersalivation and temperature instability are found in PTPS mice and human patients. Analysis of metabolites showed HPA and neopterin accumulation due to complete absence of PTPS activity, as expected (day 1 Table 2). Brain metabolites at this age were only determined for NOS due to the limited material. However, monoamine neurotransmitters must be low as inferred from measurements of 23-day-old knockouts under 'low' treatment that died despite initial treatment, and had very low or not detectable brain serotonin and dopamine levels. Although BH 4 has a central role for the function of the three NOS isoenzymes, i.e. vasorelaxation, immune response, and neurotransmission, no direct association with such pathologies in BH 4 -deficient patients has been made. Only recently, however, we found that patients revealed reduced NO metabolites in cerebrospinal fluids independent of treatment. Implications from this study are that under BH 4 -deficient conditions NOS is uncoupled and produces by-products that are neurotoxic and thus responsible for neuronal cell pathology through peroxynitrite generation (2) (37) (38). Regarding the NOS activity in our mutant mouse presented here, we found no alterations in brain nitrate and nitrite levels, independently of treatment level or age (Table 3). Furthermore, neither was the mouse brain NOS activity reduced in untreated newborn knockouts, a phenomenon that may be explained by the fact that mother milk contains high concentrations of BH 4 and that all NOS have a low K D for BH 4 binding compared to the aromatic amino acid hydroxylases (K D of 100-600 µM compared to K D of 0.2 µM for nNOS) (39). Despite lethality during the first days of life, which is not typical for BH 4 deficiency in humans, we think that this Pts knockout is a suitable animal model for studying the pathophysiology and treatment of BH 4 deficiency.
During the initial treatment study where the recommended concentration of compounds and precursors for the treatment of human patients was administered orally ('low' treatment), we learned that this therapy did not rescue the mice. Accordingly, we found no normalization of the metabolites that are followed today to control treatment in human patients, i.e. plasma phe brain biopterin, HVA, and 5-HIAA (Tables 2 and 3).
Furthermore, brain dopamine and serotonin were extremely low in the knockout mice with 'low treatment'. These neurotransmitters are below detection levels in human CSF and can thus not be determined in patients. A simple explanation for the requirement of higher doses of precursors and compounds for treatment may be the fact that mice have a much higher metabolic rate than humans. For instance, the enzymatic efficiency of PTPS is roughly ten times higher for the mouse compared to the human enzyme (k cat /K m of recombinant PTPS from mouse is 2.5 x 10 4 , and from human 2.8 x 10 3 ) (25). This hypothesis was corroborated by the fact that we could eventually rescue the animals by increasing the treatment doses .
The biochemical parameters under 'high' treatment conditions revealed that plasma phenylalanine was normalized, and biopterin was in the same range as in controls (Tables 2 and 3). Serotonin and the neurotransmitter metabolites HVA and 5-HIAA were in the subnormal range (approximately 50% of normal), whereas brain dopamine was unexpectedly low at 3% of normal (see below). Furthermore, plasma and brain neopterin remained elevated, and GTPCH was below normal activity, although hyperphenylalaninemia is expected to result in an increase of GTPCH activity. Moreover, under conditions of low BH 4 -levels, as in Pts -/mice with 'low' treatment, stimulation of GTPCH by the GTPCH-GFRP complex was expected to be even more pronounced, as the inhibitory action of BH 4 should also be diminished (4,6). Form the data presented here, we conclude that PTPS may have a direct or indirect effect on GTPCH expression or GTPCH-GFRP activity. Regarding hyperphenylalaninemia, increasing levels of oral BH 4 lead to a gradually decrease of blood phenylalanine.
Whereas the 'medium' treatment exhibited an intermediate phenylalanine level, the 'high' treatment conditions were required to completely normalize blood phenylalanine levels. A further observation that cannot be explained sufficiently at this point is that knock-out animals under 'medium' treatment consistently did not survive for more than 4 weeks, whereas under 'high' treatment none of the mutants died. The only metabolic difference we observed between these two treatment procedures was the slightly elevated plasma phenylalanine levels in the animals treated with the 'medium' dose (see Tables 2 and 3). It is unlikely that a mild HPA has such a dramatic effect on growth and development, and additional treatment studies have to be conducted to learn more about these differences.
The most remarkable observation made while treating the Pts -/mice was the consistently reduced growth starting almost from the first days of life, leading to dwarfism. Normal growth and development are largely programmed during the first weeks of postnatal life by the pituitary growth hormone (GH) and thyroxin-stimulating hormone (TSH) (40). Furthermore, somatic growth is mediated mainly by circulating IGF-I, an insulin-like hormone produced mainly in the liver but also in many other tissues. At least two examples of dwarf mice are well described: the so-called Ames and Snell dwarf mice with recessive mutations in the homeotic genes Pit-1 or Prop-1, respectively, with developmental arrest in pituitary ontogeny (40). Phenotypic characteristics are decreased growth rate post-weaning, and reduced body size of adults, having approximately one third of the weight of their normal siblings, similar to what we found in our Pts knockout mice. Further characteristics reminiscent of our dwarf mice include delayed puberty, reduction of body temperature, tendency to experience hair loss, and reduction in plasma IGF-1. This prompted us to determine these hormones in our treated mice. In contrast to the Ames and Snell dwarf mice, we found normal levels for the pituitary GH and TSH-dependent T4, indicating that the pituitary gland developed normally and can thus be excluded as the primary reason for dwarfism of treated Pts -/mice. The IGF-1 level is influenced by GH but also by the nutritional status and food intake, which in turn is regulated by the dopamine and norepinephrine levels (41,42) (43). For instance, it was reported that a knockout mouse that does not express tyrosine hydroxylase was unable to initiate feeding, an ability that can be restored by gene delivery of tyrosine hydroxylase into the striatum (32,33,44,45).
As also mentioned before, we found that the brain serotonin, and the neurotransmitter metabolites HVA, DOPAC and 5-HIAA, although slightly below normal, were not much different from control levels. This is in sharp contrast with the actual dopamine levels of 3% of normal in the brain of mice under 'high' treatment. Furthermore, the DOPAC/dopamine ratio (and the HVA/dopamine ratio), which is an index of dopamine turnover rate, was 1.8 in the wild-type and heterozygous mice group under 'medium' and 'high' treatment, but was 48 in both knockout mice groups, presumably reflecting the high turnover of the small dopamine pool (not shown). The extremely low brain dopamine together with the wealth of literature on the feeding behavior in dopaminedeficient mice mentioned above supports the assumption that abnormal (hypothalamic) neurotransmission is associated in our mice with disturbance of eating behavior. Although we did not measure daily food or water intake, we consistently observed difficulties to swallow in our knockout mice groups and conclude that control of appetite is compromised in the treated Pts -/mice and thus chronic undernutrition is responsible for low IGF-1 and dwarfism. Such feeding difficulties have been described for BH 4 -deficient patients, but so far there was no indication of growth retardation or dwarfism. Nevertheless, we tested for IGF-1 in a first study with a very small group of BH 4 -deficient patients and found specifically reduced plasma IGF-1 levels in newborns with PTPS deficiency. Although these results are only preliminary, we believe it will be important to collect more data on IGF-1 levels and to follow growth and development in human patients with BH 4 -deficiency.     Mean values and standard deviations in ng/ml are shown for normal animals (wild-type and heterozygote; n = 9) and knockouts (n = 6). The first bar is IGF-1 for both sex (f + m), whereas f is for females and m for males only. N-Acetyl-L-cysteine 4 5 2 5 5 0 ___________________________________________________ 1 The daily doses for the 'medium' and 'high' treatments were given in 2 applications, one in the morning, and one in the late afternoon. 2 The 'low' treatment corresponds to the recommended standard concentrations for treatment of human patients . 3 Decarboxylase inhibitor. 4 Antioxydant.   n.d., not detectable; n.m., not measured. 1 Days after birth. Note that from the brain of 1-day old mice not enough material could be prepared for these analyses.

LEGENDS TO FIGURES
2 See Table 1 and Materials and Methods for treatment conditions. 3 Indicated are mean values and ranges (in parenthesis); for each age group 3-4 animals were tested, except for 'low' treatment, where in some instances only two or one animal(s) were available for analysis. 4 HVA, homovanillic acid. 5 5-HIAA, 5-hydroxyindoleacetic acid.