Development of the pteridine pathway in the zebrafish, Danio rerio.

In the zebrafish, the peripheral neurons and the pigment cells are derived from the neural crest and share the pteridine pathway, which leads either to the cofactor tetrahydrobiopterin or to xanthophore pigments. The components of the pteridine pattern were identified as tetrahydrobiopterin, sepiapterin, 7-oxobiopterin, isoxanthopterin, and 2,4,7-trioxopteridine. The expression of GTP cyclohydrolase I activity during the first 24-h postfertilization, followed by 6-pyruvoyl-5,6,7,8-tetrahydropterin synthase and sepiapterin reductase, suggest an early supply of tetrahydrobiopterin for neurotransmitter synthesis in the neurons and for tyrosine supply in the melanophores. At 48-h postfertilization, sepiapterin formation branches off the de novo pathway of tetrahydrobiopterin synthesis. Sepiapterin, via 7,8-dihydrobiopterin and biopterin, serves as a precursor for the formation of 7-oxobiopterin, which may be further catabolized to isoxanthopterin and 2,4,7-trioxopteridine. Neither 7, 8-dihydrobiopterin nor biopterin is a substrate for xanthine oxidoreductase. In contrast, both of these compounds are oxidized at C-7 by a xanthine oxidase variant form, which is inactivated by KCN, but is insensitive to allopurinol. The oxidase and the dehydrogenase form of xanthine oxidoreductase as well as the xanthine oxidase variant have specific developmental patterns. It follows that GTP cyclohydrolase I, the formation of sepiapterin, and the xanthine oxidoreductase family control the pteridine pathway in the zebrafish.

The zebrafish has a number of features that have led to its widespread use as a model system for genetic analysis and for the study of vertebrate development (1). Several studies have been reported on the staging, cell fate determination, lineage determination, and patterning for the development of the peripheral neurons and the pigment cells in this animal. Both of these types of cells are derived from the neural crest (2)(3)(4). Large scale screens for embryonic and early larval zebrafish mutations have enabled the analysis of different aspects of neuronal and of pigment development and the identification of the genes specifically required for them (5)(6)(7)(8). In the zebrafish, some of the pigments are believed to be pteridine-related, and differences in their biosynthesis were suggested to occur in several mutants (6).
The de novo biosynthesis of pteridines begins with GTP and results in the formation of H 4 biopterin. 1 The first committed step of this pathway is catalyzed by GTP cyclohydrolase I (EC 3.5.4.16) and yields H 2 neopterin triphosphate (9). 6-Pyruvoyl-H 4 pterin synthase (EC 4.6.1.10) eliminates the phosphate groups from this and catalyzes an intramolecular reaction, yielding 6-pyruvoyl-H 4 pterin (10). Both keto groups of this unstable intermediate are reduced by sepiapterin reductase (EC 1.1.1.153) via formation of 6-(1Ј-hydroxy-2Ј-oxopropyl)-H 4 pterin, isomerization to 6-lactoyl-H 4 pterin, and final reduction of this 1Ј-oxo compound to H 4 biopterin; the necessary hydride equivalents come from NADPH (11; pathway reviewed in Ref. 12). H 4 biopterin serves as the electron donor for the hydroxylation of the aromatic amino acids phenylalanine, tyrosine, and tryptophan and thus is the essential cofactor for the synthesis of the neurotransmitters, catecholamine and serotonin (reviewed in Ref. 13). H 4 biopterin is therefore synthesized in the neurons, and its subsaturating levels, in some cases, control neurotransmitter synthesis (14,15). H 4 biopterin is also an essential component of nitric oxide synthase (reviewed in Refs. 16,17). The diversity of nitric oxide function as a diffusible intercellular messenger in fishes is, however, not yet explained in such detail as in mammals (18) and not yet analyzed in the zebrafish.
During the development of vertebrates, multipotent neural crest cells give rise to peripheral neuronal cells as well as to pigment cells (27;reviewed in Ref. 28). Although these two types of cells vary widely in their phenotype and cellular function, it nevertheless seems clear that both of them must express the enzymes that catalyze the de novo synthesis of pteridines from GTP. So far, the numerous genetic and developmental studies that address neuronal defects, changes in pigment cell pattern, or both have focussed on the description of the phenotype. The aim of the present work, therefore, was to analyze the pattern of how pteridines occur in the zebrafish, to elucidate the biosynthetic pathway leading to them, and thus to identify committing steps in the biosynthesis of neurotransmitters, of melanin, and of xanthophore pigments during the development of the organism.

EXPERIMENTAL PROCEDURES
Animals-Wild type zebrafish of the AB background were raised in the zebrafish facility of the Institute of Mammalian Genetics, Neuherberg, and embryos were obtained from natural spawning. They were grouped according to number postfertilization hours at 28.5°C. The samples were collected in batches of 10 -20 animals, immediately frozen, and kept at Ϫ85°C until analysis.
Reagents-Chemicals used for the determination of enzymatic activities and for protein and pteridine determination were obtained as described previously (29,30). Pterins were from Schircks (Jona, Switzerland). 2,4,7-Trioxopteridine was prepared by a modification of a previously described method (31) (see below). A sample of the same compound, as well as 7-oxo-6-carboxypterin and 7-oxobiopterin isolated from carp skin were donated by Dr. M. Nakagoshi (Kitasato). Construction of the plasmids used for expression of murine recombinant GTP cyclohydrolase I and for murine sepiapterin reductase followed the procedures described earlier (11,32). Recombinant 6-pyruvoyl-H 4 pterin synthase was a gift from Dr. A. Bacher (Mü nchen). Milk xanthine oxidoreductase was obtained from Sigma, alkaline phosphatase from Roche Molecular Biochemicals, MicroSpin G25 columns from Amersham Pharmacia Biotech, Sep-Pak cartridges from Waters Corporation, and cellulose thin layer chromatography plates from Merck.
Identification of the Pteridine Pattern-The samples (batches of 50 animals, 96-to 120-h postfertilization) were extracted by homogenization with a Wheaton Teflon homogenizer in 100 ml of Tris buffer (50 mM, pH 6.9) and centrifuged at 13,000 ϫ g for 10 min. The extract was oxidized by acidic iodine, deproteinized by trichloroacetic acid, and purified by cation-exchange chromatography, as described previously. The lyophilized material was prefractionated by reverse-phase high pressure liquid chromatography (HPLC), using 3% methanol in water as the mobile phase, and a fluorescence detector, as described before (29,30). The fractions were concentrated in vacuo.
Three main fractions were obtained and identified using a combination of physical and chemical methods. HPLC analysis using a Shimadzu model RF 535 fluorescence detector was carried out on reversephase columns using five different solvent systems, the compositions of which are detailed below (see the legend of Table I). Mass spectrometric analysis was carried out using a Micromass LCT instrument with electrospray ionization and time-of-flight analysis. The source temper-ature was 30°C, and the cone voltage was 30 V. Fractions I and II were dissolved in 30 l of 0.1 M ammonium hydroxide, and 30 l of 15% methanol in water, respectively, and injected onto a C-18 reverse-phase HPLC column leading into the mass spectrometer. For further identification of biopterin and of 7-oxobiopterin in Fractions II and III, respectively, the fractions were oxidized by alkaline permanganate, and the resulting pterin-6-carboxylic acids were separated by HPLC and fluorometrically detected as described previously (34).
Sepiapterin was identified by its enzymatic reduction to H 2 biopterin. A centrifuged homogenate (Tris buffer, 50 mM, pH 6.9) of 40 animals (120-h postfertilization) was concentrated by loading onto Sep-Pak cartridges as described before (35), desorbed with methanol, concentrated in a vacuum, spotted on cellulose thin layer plates, and developed with butanol/acetic acid/water (4:1:5, v/v). The yellow fluorescent spots (R F ϭ 0.17) were extracted with Tris buffer (50 mM, pH 7.9), centrifuged, and incubated for 40 min with NADPH (200 M) and recombinant sepiapterin reductase for 40 min. After acidic iodine oxidation of the reaction product H 2 biopterin, the sample was analyzed by HPLC in solvent systems I and II (see legend of Table I).
Quantitative Determination of the Pteridine Pattern during Development-Batches of 10 -20 animals were extracted, oxidized, deproteinized, and prepurified by cation-exchange chromatography as described above. Because isoxanthopterin is only partially retained and also 7-oxobiopterin and 2,4,7-trioxopteridine are only partially retained, the lyophilized water washes were separately monitored and their amounts combined. For quantitative determination by HPLC analysis, the solvent systems I and II (see legend of Table I) were used. Due to the different fluorescence characteristics of biopterin and, conversely, of the pteridines substituted at the 7-position, the samples were either monitored at two different wavelength settings (350-nm excitation/450-nm emission and 340-nm excitation/420-nm emission) or the relative emission characteristics were determined for each series, and the samples were then monitored at only one wavelength setting (350-nm excitation/ 450-nm emission) and the relevant factors were applied for calculations. For the determination of sepiapterin, the samples were homogenized in 1 ml of Tris buffer (50 mM, pH 6.8), centrifuged, and then adsorbed onto reverse-phase C-18 Sep-Pak cartridges (35). After washing, desorption with 3 ml of methanol, and concentration by vacuum, sepiapterin was determined by its absorption at 440 nm. The standard curve was linear in the range of 0 -15 nmol/ml.
Enzyme Assays-For all enzyme assays, batches of 10 -20 animals were homogenized in Tris buffer (50 mM, pH 8.0) as described above. The centrifuged homogenates (13,000 ϫ g, 15 min) were passed through MicroSpin G25 columns. For determination of GTP cyclohydrolase I activity, the reaction product H 2 neopterin triphosphate was oxidized to neopterin triphosphate by acidic iodine solution. After reduction of excess iodine by ascorbic acid and dephosphorylation with alkaline phosphatase (0.8 unit/200 l), neopterin was determined by HPLC using solvent systems I and II (see legend of Table I). For determination of 6-pyruvoyl-H 4 pterin synthase, H 2 neopterin triphosphate was first generated by incubation of GTP with recombinant murine GTP cyclohydrolase I. To determine the activity of the synthase, the extract containing it (plus Mg 2ϩ , to achieve a final concentration of 10 mM) was added, and the unstable intermediate 6-pyruvoyl-H 4 pterin was converted to H 4 biopterin by the addition of recombinant murine sepiapterin reductase and NADPH (200 M). After acidic iodine oxidation, biopterin was separated by HPLC (36). The activity of sepiapterin reductase was determined by reduction of sepiapterin to dihydrobiopterin with NADPH, acidic iodine oxidation, and HPLC separation of the resultant biopterin. The detailed conditions for all enzyme assays were as described previously (30). For preparation of 6-pyruvoyl-H 4 pterin as a potential substrate for the xanthine oxidoreductase variant, H 2 neopterin triphosphate was first generated by incubation of GTP with recombinant murine GTP cyclohydrolase I, and EDTA was quenched by addition of Mg 2ϩ as described above; the reaction mixture was incubated with recombinant 6-pyruvoyl-H 4 pterin synthase for 20 min and then immediately used. Due to the endogenous sepiapterin reductase activity of zebrafish homogenates and the additional function of this enzyme as an isomerase (11,37), the intermediates 6-lactoyl-H 4 pterin and 6-(1Ј-hydroxy-2Ј-oxopropyl)-H 4 pterin are transiently formed also, so that these could also be considered as possible precursors for 7-oxobiopterin formation. The assay for xanthine oxidoreductase activity used the conversion of pterin to isoxanthopterin, which was then separated by HPLC (solvent system IV; see legend of Table I) and fluorometrically detected (340-nm excitation/410-nm emission) (38,39). The instrument was standardized by measurements with the product (isoxanthopterin) and the substrate. For the determination of the oxidase activity, the reaction mixture (100 l) contained the extract (30 l), pterin (50 M), and air-saturated Tris buffer (50 mM, pH 8.0). For the determination of combined dehydrogenase plus oxidase activity, NAD (400 M) was included. The samples were incubated for 40 min at 37°C, and the reaction was stopped by addition of 2 N trichloroacetic acid (20 l). Protein was estimated by the Coomassie dye binding reagent (Bio-Rad protein assay reagent) according to the manufacturer's instructions.

Identification of the Pteridine Pattern-
The extracts from 4to 5-day-old zebrafish, which had been purified by cation-exchange chromatography and then prefractionated by reversephase HPLC with 3% methanol, yielded three major fractions, labeled I through III. Each of these fractions was investigated separately, and the pteridines in them were identified as described below.
Fraction I eluted at 9.8 min. On further reverse-phase HPLC analysis with a fluorescence detector, it showed one peak that coeluted with authentic isoxanthopterin in all of five different solvent systems (Table I). Its maximum fluorescence emission at 410 nm, and its UV spectra in both 0.1 M HCl and 0.1 M NaOH were identical with those of authentic isoxanthopterin. The two strongest peaks in its electrospray mass spectrum occurred at m/z 180 (M ϩ H ϩ ) and 218 (M ϩ K ϩ ) and were identical with the two strongest peaks of authentic isoxanthopterin measured under the same conditions. Fraction II eluted at 11.0 min. On further reverse-phase HPLC analysis with a fluorescence detector, it showed two peaks in four out of five solvent systems tested (Table I), and these were identified as 2,4,7-trioxopteridine and biopterin. First, the two peaks had the same retention times as authentic samples of 2,4,7-trioxopteridine and biopterin. Second, the excitation/emission wavelengths of the first fluorescence peak were shifted from 350/450 nm to 340/420 nm, reaching 3-fold enhancement. This is typical of 7-oxopteridines (40 -42). Third, the mass spectrum of the first fluorescence peak showed a strong ion at m/z 203 (M ϩ Na ϩ ). Authentic 2,4,7-trioxopteridine, measured under the same conditions, also shows a strong ion at m/z 203. Fourth, using liquid chromatography-mass spectrometry, the compound in the second fluorescence peak was found to have both UV and mass spectra identical with those of authentic biopterin. The mass spectrum showed m/z peaks at 238 (M ϩ H ϩ ), 260 (M ϩ Na ϩ ), and 276 (M ϩ K ϩ ). Fifth, alkaline permanganate oxidation of the second peak yielded pterin-6-carboxylic acid (Fig. 1A). Alkaline iodine oxidation of zebrafish extracts, which decomposes H 4 biopterin (43), left no biopterin. This demonstrated that the total biopterin fraction exists as H 4 biopterin.
Fraction III eluted at 13.9 min. It showed a 1.5-fold increase in fluorescence intensity at 420 nm as compared with 450 nm, indicating a 7-oxopteridine, and HPLC analysis in solvent sys-tems I-IV (Table I) suggested its identity with 7-oxobiopterin. This was confirmed by alkaline permanganate oxidation of Fraction III to 6-carboxy-7-oxopterin (Fig. 1B).
The yellow pterin, isolated from the Sep-Pak cartridges, had an absorption maximum at 340 nm in 0.1 M NaOH and at 410 nm in 0.1 M HCl, as is characteristic for sepiapterin (44). Reduction of it with sepiapterin reductase plus NADPH yielded H 2 biopterin, which was oxidized by acidic iodine to biopterin, identified by HPLC analysis with solvent systems I and II ( Fig. 1C and Table I) Development of the Pteridine Pattern-The individual pteridines, which constitute the final pattern of pigments in the xanthophores, were found to have different developmental profiles (Fig. 2). Both H 4 biopterin and sepiapterin were the first ones to form, appearing within 24-h postfertilization. H 4 biopterin rapidly increased to a maximum after 72 h, and finally stayed constant around 7 pmol per animal. In contrast, sepiapterin continued to increase during the following period up to a 10-fold level. Formation of the three 7-oxopteridines (7-oxobiopterin, isoxanthopterin, and 2,4,7-trioxopteridine) began after an initial time lag of 24 h, but their concentrations then increased rapidly. This indicated that the de novo biosynthetic pathway must first provide the substrate for the subsequent formation of the 7-oxo derivatives.
Development of H 4 biopterin Synthesizing Enzymes-The three enzymes that catalyze the biosynthesis of H 4 biopterin from GTP jointly start to develop during the first 24-h postfertilization (Fig. 3A). This agrees with the early increase in H 4 biopterin levels (Fig. 2). However, their activities diverge after 72 h: GTP cyclohydrolase I activity reaches a maximum after 72 h and then declines, whereas both 6-pyruvoyl-H 4 pterin synthase and sepiapterin reductase activities continue to increase after 72 h.
Pterin Metabolism by Xanthine Oxidoreductase and a Xanthine Oxidoreductase Variant-Xanthine oxidoreductase efficiently converts pterin into isoxanthopterin (38,45). This enzyme was therefore a likely candidate for the generation of the 7-oxopterins from intermediates that are synthesized along the route of the de novo pathway in the zebrafish. The enzyme exists as two interconvertible forms, xanthine dehydrogenase and xanthine oxidoreductase, which use either NAD ϩ (the dehydrogenase) or molecular oxygen (the oxidase) as the electron acceptor. In the same way as purified milk xanthine oxidoreductase, extracts of zebrafish (72-h postfertilization) were able to catalyze the oxygenation of pterin to isoxanthopterin. The yield was 83.9 Ϯ 13.1 pmol/mg/min (O 2 as electron acceptor) and 124 Ϯ 20.8 pmol/mg/min (NAD ϩ as electron acceptor).
The development of the pteridine pattern (Fig. 2) suggests that at approximately 48-h postfertilization there is a switch from production of H 4 biopterin to production of the 7-oxopteri-TABLE I HPLC analysis of the zebrafish pteridine pattern Extracts of zebrafish (96 -120 h postfertilization) were prepurified as described under "Experimental Procedures" and prefractionated by reverse-phase HPLC with 3% methanol. The fractions (I-III) were then analyzed in five solvent systems. The compositions of the solvent systems were: I, 2-propanol 0.4%, triethylamine 1%, adjusted to pH 7.0 with H 3 PO 4 ; II, 3% methanol, 1% acetonitrile, 0.02% phosphoric acid; III, 3% methanol, 1% acetonitrile; IV, K 2 HPO 4 3.5 mM, adjusted to pH 7.2 with H 3 PO 4 ; V, 5% methanol, NaH 2 PO 4 100 mM, adjusted to pH 3 with H 3 PO 4 . The flow rate was 1 ml/min; the numbers refer to the retention time (minutes). dines. In search of the actual precursor from which 7-oxobiopterin is formed from an intermediate of the de novo pathway, zebrafish extracts were incubated with H 4 biopterin and with 6-pyruvoyl-H 4 pterin. The yield of 7-oxobiopterin from these substrates, using either oxygen or NAD ϩ as electron acceptor, was Ͻ0.5%. In the 6-pyruvoyl-H 4 pterin experiment, only H 4 biopterin (30 Ϯ 4.5 pmol/mg/min) but no 7-oxobiopterin was formed, when NADPH (200 M) was included. This was to be expected, because it is known that endogenous sepiapterin reductase activity (Fig. 3A), coupled to the action of the enzyme as an isomerase (37), can convert 6-pyruvoyl-H 4 pterin into H 4 biopterin, through the intermediates 6-lactoyl-H 4 pterin and 6-(1Ј-hydroxy-2Ј-oxopropyl)-H 4 pterin. This shows that the latter two compounds as well as 6-pyruvoyl-H 4 pterin and H 4 biopterin do not serve as substrates for 7-oxobiopterin formation.
Although H 2 biopterin is not believed to be an intermediate in the de novo route to H 4 biopterin, it is nevertheless formed from sepiapterin. The reaction is also catalyzed by sepiapterin reductase and requires NADPH (46). The resultant H 2 biopterin can be oxidized to biopterin by dihydropterin dehydrogenase (47) so that both H 2 biopterin and biopterin become candidates for the biosynthetic precursor of 7-oxobiopterin. Although it was found, as expected, that sepiapterin (40 M), in the absence of NADPH did not serve as a direct substrate, inclusion of NADPH (200 M) in the experiment resulted in the formation

FIG. 1. Reverse-phase HPLC of the pterin product formed after alkaline permanganate oxidation of Fractions II and III and after enzymatic reduction of the yellow pigment by recombinant sepiapterin reductase.
For preparation of Fractions II and III, the extracts of 4-to 5-day-old zebrafish were subjected to acidic iodine oxidation, prepurified by Dowex H ϩ exchange chromatography, and prefractionated by reverse-phase HPLC with 3% methanol as eluent. For preparation of the yellow pigment, the extract was concentrated by Sep-Pak cartridges and then separated by cellulose thin layer chromatography. A, permanganate oxidation product of Fraction II. The compound eluting at 12.0 min with solvent V (see legend of Table I) is identified as 6-carboxylic acid. A1, elution profile of the sample; A2, elution profile of synthetic 6-carboxylic acid. Both show a 140% relative fluorescence at 350-nm excitation/450-nm emission as compared with 340-nm excitation/420-nm emission. B, permanganate oxidation product of Fraction III. The compound eluting at 13.5 min with solvent V (see legend of Table I) is identified with 7-oxo-6-carboxylic acid. B1, elution profile of the sample; B2, elution profile of synthetic 7-oxo-6carboxylic acid. Both show a 150% relative fluorescence at 340-nm excitation/420-nm emission as compared with 350-nm excitation/ 450-nm emission. C, enzymatic reduction of the yellow pigment. It was reduced to H 2 biopterin by recombinant sepiapterin reductase and NADPH and then oxidized by acidic iodine solution. The compound eluting at 7.4 min with solvent II (see legend of Table I)  Extracts were subjected to acidic iodine oxidation, prepurified, and analyzed by reverse-phase HPLC. Sepiapterin was determined in Tris buffer extracts (pH 6.9) after concentration with Sep-Pak cartridges by its absorption at 440 nm.

FIG. 3. Activities of the enzymes involved in de novo H 4 biopterin synthesis of xanthine oxidase and the xanthine oxidase variant during the development of the zebrafish.
A, enzymes of the de novo H 4 biopterin synthesis. f, GTP cyclohydrolase I (product: H 2 neopterin triphosphate); q, 6-pyruvoyl-H 4 pterin synthase (product: H 4 biopterin); OE, sepiapterin reductase (product: H 2 biopterin). Extracts were assayed as described under "Experimental Procedures." The products were subjected to acidic iodine oxidation and analyzed by reverse-phase HPLC. B, f, xanthine oxidase, product: isoxanthopterin; q, xanthine oxidase ϩ dehydrogenase, product: isoxanthopterin; OE, xanthine oxidase variant, product: 7-oxobiopterin. Extracts were assayed for xanthine oxidoreductase with pterin as substrate and O 2 (oxidase activity) or NAD ϩ as electron acceptor (oxidase ϩ dehydrogenase activity) and for the xanthine oxidase variant with biopterin as substrate and O 2 as electron acceptor. Isoxanthopterin and 7-oxobiopterin were analyzed by reverse-phase HPLC. of 7-oxobiopterin. Furthermore, it was found that the known inhibitor of sepiapterin reductase, N-acetylserotonin (48) not only blocked the formation of H 2 biopterin by endogenous sepiapterin reductase, but also inhibited the formation of 7-oxobiopterin. In both of these reactions, half-maximal inhibition occurred at 0.3 M (data not shown). These results show that formation of 7-oxobiopterin depends upon the reduction of sepiapterin and suggest that either H 2 biopterin or biopterin serve as the direct precursors. In fact, it was found that with H 2 biopterin as a substrate, zebrafish extracts (72-h postfertilization) yielded 8.8 Ϯ 1.5 pmol of 7-oxobiopterin/mg/min. With biopterin, the yield of 7-oxobiopterin was 52.6 Ϯ 5.5 pmol/mg/ min (Fig. 4); inclusion of NAD ϩ had no effect.
It was also found, however, that purified xanthine oxidoreductase was unable to oxidize either H 2 biopterin or biopterin to 7-oxobiopterin, neither with O 2 nor with NADPH ϩ as electron acceptor (data not shown). Allopurinol is a potent inhibitor of xanthine oxidoreductase (49). Although this enzyme is almost completely inhibited at concentrations of 1.25 mM, the formation of 7-oxobiopterin from H 2 biopterin or biopterin by zebrafish extract was found to be unaffected by allopurinol at concentrations up to 1 mM (Fig. 5). These data indicate that the enzyme that catalyzes the formation of 7-oxobiopterin from H 2 biopterin or biopterin in the zebrafish, is related to but is not identical to xanthine oxidoreductase. In keeping with this, it was found that oxidation of H 2 biopterin or biopterin to 7-oxobiopterin by zebrafish extract was completely suppressed by cyanide (5 mM), using either O 2 or NAD ϩ as electron acceptor (data not shown).All members of the xanthine oxidase family of enzymes are known to be inactivated by cyanide (50).
Development of Xanthine Oxidoreductase and the Xanthine Oxidoreductase-related Enzyme-The dehydrogenase form of xanthine oxidoreductase can be converted reversibly to the oxidase by formation of disulfide bridges from vicinal residues, or by irreversible splitting through limited proteolysis (51,52), but it is not yet decided whether both entities may exist as such in the cell. We have therefore analyzed separately the activities of xanthine oxidoreductase and of xanthine dehydrogenase as well as the development of the xanthine oxidoreductase-related enzyme in the zebrafish. Fig. 3B demonstrates that the activities of xanthine oxidoreductase and of xanthine dehydrogenase diverge after 48-h postferilization and that the dehydrogenase function dominates after 72 h. Additions to the assay of dithiothreitol (1 mM), the protease inhibitor phenylmethylsulfonyl fluoride (200 M), or the metalloprotease inhibitor EDTA (2.5 mM), separately or together, did not change the developmental pattern (data not shown). This indicates that the activities of the dehydrogenase and of the oxidase form are not caused by interconversion of both enzyme forms during the assay. Both forms rather appear to be present in the cell in a stage-specific fraction of total oxidoreductase.
During the first 72-h postfertilization, the activity of the xanthine oxidoreductase-related enzyme develops more slowly than the activities of both xanthine oxidoreductase forms (Fig.  3B). Around the hatching period, the increasing slope in its activity results in almost the same activity levels at 120-h postfertilization.

DISCUSSION
In this study we have identified H 4 biopterin, sepiapterin, 7-oxobiopterin, 2,4,7-trioxopteridine, and isoxanthopterin as components of the pteridine pattern in zebrafish and analyzed their biosynthesis during the first 120-h postfertilization. At this time, the embryonic arrangement of the pigment cells is essentially established (5). Furthermore, our results suggest that the pteridine pattern involves two biosynthetic routes (Fig. 6). The first one leads from GTP to H 4 biopterin (reviewed in Ref. 12), providing H 4 biopterin as the essential cofactor for neurotransmitter synthesis in the developing neurons (13,15) and for phenylalanine hydroxylation in the melanophores (21). This de novo synthetic pathway is well understood in Drosophila (53) and in mammalian cells (reviewed in Ref. 12), and it proceeds via tetrahydropterin intermediates once the unstable 6-pyruvoyl-H 4 pterin is formed from H 2 neopterin triphosphate.
The second route starts from sepiapterin, which accumulates during the first 24-h postfertilization. The sepiapterin is reduced by sepiapterin reductase to H 2 biopterin, which can be oxidized to biopterin by a dihydrobiopterin oxidase. This en-  Table I), is identified as 7-oxobiopterin. A, elution profile of the sample at t o ; B, elution profile of the sample after 40 min of incubation. The substrate fraction (biopterin, elution at 8.3 min) shows a 150% relative fluorescence at 350-nm excitation/450-nm emission as compared with 340-nm excitation/420-nm emission, whereas the product eluting at 11.4 min (7-oxobiopterin) shows a 150% relative fluorescence at 340-nm excitation/420-nm emission as compared with 350-nm excitation/450-nm emission. zyme was isolated from Drosophila (47), but its occurrence in vertebrates has not been reported so far, nor are methods yet available that could enable studies of its synthesis in crude extracts. It is likely that either H 2 biopterin or biopterin (which gives a 16-fold higher yield) is oxidized at position C-7 by a xanthine oxidoreductase variant. It is not clear at what stage the 3-carbon side chain is lost in the formation of the isoxanthopterin and the 2,4,7-trioxopteridine. It is also uncertain whether or not this side chain cleavage is enzymatically controlled (reviewed in Ref. 54). Possibly, the side chain is lost prior to oxidation at position C-7, because both the pterin and lumazine are well defined substrates for xanthine oxidoreductase (42,45); this depends upon prior cleavage of the C-6 side chain and (in the case of lumazine) upon deamination at the C-2 position. The properties of pterin deaminase (EC 3.5.4.11) from eukaryotes such as Dictyostelium discoideum, or rat liver, are poorly defined as yet (for review see Ref. 54) The results suggest that the overall process of pteridine synthesis in the developing zebrafish embryo is controlled by three reactions. First, an initial surge of GTP cyclohydrolase I activity promotes a rapid increase in the synthesis of the H 4 biopterin cofactor. After 72-h postfertilization, however, GTP cyclohydrolase I activity does not correlate with the onset of rapid production of pteridine pigments. In mammalian cells, various cytokines and growth factors regulate the activity of GTP cyclohydrolase I at the transcriptional or post-transcriptional level (55)(56)(57), through post-translational modification by hyperphosphorylation (58) or by association with a feedback regulatory protein (59). Research is under way into the regulation of GTP cyclohydrolase I activity during the development of the zebrafish and in particular to explain why this activity starts to decrease at around 72-h postfertilization, which coincides with the production of xanthophore pteridine pigments.
Second, the build up of sepiapterin provides the key precursor for the 7-oxopteridines in xanthophores. The precise origin of the sepiapterin remains obscure. In Drosophila melanogaster it has been proposed that sepiapterin may arise by air oxida-tion of the 6-lactoyl-H 4 pterin, which lies on the H 4 biopterin pathway (53,60). Another possibility is that it could be formed from 6-pyruvoyl-H 4 pterin or 6-lactoyl-H 4 pterin under enzymatic control, e.g. by a dismutase-type reaction, although this is still hypothetical. Sepiapterin assumed importance in earlier work because of its possible role in a salvage pathway to H 4 biopterin in mammalian cells (61). Here also, however, its origin remains obscure.
Third, a xanthine oxidase variant appears to control the formation of 7-oxobiopterin, which then may give rise to isoxanthopterin and 2,4,7-trioxopteridine. An enzyme related but not identical to xanthine oxidase has already been observed in carp skin, and biopterin has been suggested as its substrate (62). This variant form is grouped with the xanthine oxidoreductase family because of its inactivation by cyanide, which involves elimination of its sulfur in the vicinity of Mo 6ϩ (50). The xanthine oxidoreductase family of enzymes has the same overall general structure but different substrate specificities (reviewed in Ref. 63). The insensitivity of the zebrafish xanthine oxidoreductase variant to allopurinol, whose "suicide" product allo-xanthine binds to the active site (49), is consistent with its different substrate specificity. The three related enzymes, namely the oxidase and dehydrogenase form of xanthine oxidoreductase, and the xanthine oxidase variant, all have different developmental patterns. This suggests that expression of different forms of xanthine oxidoreductase in xanthophores is stage-specific. The xanthine oxidoreductase gene is characterized by high rates of exon multiplication during phylogeny, the functional meaning of which is unknown. The expression is tissue-and cell-specific, and it is regulated by different stimuli (reviewed in Ref. 63). Combined genetic and biochemical approaches in the zebrafish may help to elucidate gene structure and regulation of the xanthine oxidoreductase family.
In summary, our results suggest that GTP cyclohydrolase I, the xanthine oxidoreductase family, and an as yet unexplained reaction leading to sepiapterin, control the pteridine pathway FIG. 6. Proposed pathway leading to the zebrafish pteridines. The sequential action of GTP cyclohydrolase I, 6-pyruvoyl-H 4 pterin synthase, and sepiapterin reductase catalyzes the de novo synthesis of the H 4 biopterin cofactor from GTP. Sepiapterin takes a pool position; it is accumulated as pigment and, after reduction of the C-6 side chain by sepiapterin reductase, additionally gives rise to 7-oxobiopterin, involving oxygenation at C-7 by a xanthine oxidase variant. Isoxanthopterin and 2,4,7-trioxopteridine may arise from either 7-oxobiopterin or from intermediates on its pathway by a series of enzymatic and nonenzymatic reactions such as deamination, side chain cleavage, and oxygenation of the pteridine core by xanthine oxidoreductase. 1, GTP cyclohydrolase I; 2, 6-pyruvoyl-H 4 pterin synthase; 3, sepiapterin reductase; 4, H 2 pterin oxidase; 5, xanthine oxidase variant; 6, pterin deaminase; 7, xanthine oxidoreductase; 8, nonenzymatic side chain cleavage; asterisk, hypothetical enzymatic reaction.
in zebrafish embryos. This may allow for a candidate gene approach to explain the mechanisms that lead to pigment mutations by loss of pigment synthesis and to those mutations that affect neuronal function.