An aldehyde dehydrogenase present at high levels in the dorsal retina of the embryonic and adult mouse was identified as the isoform AHD-2 known to oxidize retinaldehyde to retinoic acid. Comparative estimates of retinoic acid levels with a reporter cell line placed the retinas among the richest tissues in the entire body of the early embryo; levels in ventral retina, however, exceeded dorsal levels. Retinoic acid synthesis from retinaldehyde in the dorsal pathway was less effective than the ventral pathway at low substrate levels and more effective at high levels. The dorsal pathway was preferentially inhibited by disulfiram, while ventral synthesis was preferentially inhibited by p-hydroxymer-curibenzoate. When protein fractions separated by isoelectric focusing were analyzed for retinoic acid synthesizing capacity by a zymography-bioassay, most of the synthesis in dorsal retina was found to be mediated by AHD-2, and ventral synthesis was mediated by dehydrogenase activities distinct in charge from AHD-2. Postnatally, levels of highest retinoic acid synthesis shifted from ventral to dorsal retina. In the adult retina, the dorsal pathway persisted, but the preferential ventral pathway was no longer detectable. Our observations raise the possibility that retinoic acid plays a role in the determination and maintenance of the dorsoventral axis of the retina, and that the morphogenetically significant asymmetry here lies in the spatial arrangement of synthetic pathways.

An abundant protein first noticed in the dorsal embryonic retina was identified as a cytosolic aldehyde dehydrogenase (McCaffery et al., 1991). Its distribution in the mouse is summarized in Fig. 1: it appears early on embryonic day 9 (E9) in the dorsal part of the eye vesicle, and it maintains a dorsal localization throughout development up to the adult retina. Aldehyde dehydrogenases are a group of enzymes of similar molecular weight which differ in a range of characteristics including subcellular distribution, charge, relative substrate preference and inhibitor susceptibility (Lin-dahl and Hempel, 1990; Manthey et al., 1990). The enzyme in dorsal retina resembled the isoform AHD-2, which has been shown as the major nicotinamideadenine dinucleotide (NAD) linked enzyme in the adult mouse liver capable of retinoic acid synthesis from retinaldehyde (Lee et al., 1991). In other species, this isoform is known as El or ALDH-1 (Greenfield and Pietruszko, 1977; Lindahl and Hempel, 1990). Here we show (1) that the enzyme in dorsal retina is identical to AHD-2; (2) that it generates retinoic acid in vitro and probably also in vivo; (3) that there are other, more powerful routes of retinoic acid synthesis in the retina; and (4) that retinoic acid levels and synthetic pathways are asymmetrically arranged in the dorsoventral axis of the retina.

Fig. 1.

Sections through the eyes of mice at E9 (A), E10 (B), E12 (C), E16 (D) and adult age (E), labeled with aldehyde dehydrogenase antisera (Lindahl and Evces, 1984a,b; Russo and Hilton, 1988). The plane of the sections runs parallel to the dorsoventral axis of the retina. The aldehyde dehydrogenase appears early on E9 in the dorsal eye vesicle and remains dorsal except for a transient minor focus in ventral retina in the later embryo as seen here for E16. Scales: A-D, 200 μ m;E, 500 μ m.

Fig. 1.

Sections through the eyes of mice at E9 (A), E10 (B), E12 (C), E16 (D) and adult age (E), labeled with aldehyde dehydrogenase antisera (Lindahl and Evces, 1984a,b; Russo and Hilton, 1988). The plane of the sections runs parallel to the dorsoventral axis of the retina. The aldehyde dehydrogenase appears early on E9 in the dorsal eye vesicle and remains dorsal except for a transient minor focus in ventral retina in the later embryo as seen here for E16. Scales: A-D, 200 μ m;E, 500 μ m.

Retinoic acid is known to affect the development of axial patterning in the embryo (Brockes, 1989; Tabin, 1991): application of retinoic acid to the wing bud of the chick embryo mimics the effects of the zone of polarizing activity (Noji et al., 1991; Tickle et al., 1982; Wanek et al., 1991), and along the body axis it causes a dose-dependent reduction in anterior neural and mesodermal structures (Durston et al., 1989; Ruiz i Altaba and Jessell, 1991). Its action is mediated by nuclear receptors, which are members of the steroid-thyroid hormone receptor family of transcription regulators. Three subtypes of retinoic acid receptors have been identified in mammals, the RAR α, RARβ and RAR γ (Benbrook et al., 1988; Brand et al., 1988; Giguère et al., 1987; Krust et al., 1989; Petkovich et al., 1987; Zelent et al., 1989), which show distinct patterns of distribution in the embryo (Dollé et al., 1989, 1990; Osumi-Yamashita et al., 1990; Ruberte et al., 1991,1990). The retinoic acid-activated RARs direct transcription by binding to retinoic acid responsive elements (RARE) in the promotor regions of certain genes including several which contain the DNA-binding homeobox and Pax (paired-box) motifs. Retinoic acid thus represents a universal effector molecule capable of activating cascades of gene expression. Pattern formation, for instance in the limb bud, is thought to be achieved through a series of events: following the retinoic acid-sensitive axial determination, overt differentiation is initiated and preceded by prepatterns of specific retinoic acid receptors, binding proteins and homeobox transcripts (Brockes, 1989; Dollé et al., 1989; Izpisua-Belmonte et al., 1991; Maden et al., 1989; Tabin, 1991).

All mice were from an outbred colony. The embryos were staged according to Theiler (1972). Tissues were dissected in tissue culture medium, washed in balanced salt solution, and either directly plated onto the reporter cells or frozen in liquid nitrogen and stored at —80°C. All embryos used for the dorsoventral comparisons were between E12.5 and 14.5, mostly between E13 and 14. For immunohistochemistry mice were fixed and processed as described previously (McCaffery et al., 1991), and the sections were labeled with aldehyde dehydrogenase antisera (Lindahl and Evces, 1984a, 1984b; Russo and Hilton, 1988) in combination with a fluorescent detection method.

For the isoelectric focusing zymogram (Fig. 2), 191 retinas from E13-15 mice were homogenized in 20 mM triethanol-amine-HCl buffer (TEA-HC1) at pH 7.4 in the presence of 1 mM dithiothreitol (DTT), 0.1 mM ethylene glycol-(bis aminoethyl ether) tetra-acetic acid (EDTA), 10 mg/ml aprotinin and 1 mM phenylmethyl sulfonyl fluoride (PMSF). The 105,000g/45 minute supernatant was run in parallel with a partially purified AHD-2 preparation obtained through 5’-AMP-Sepharose column chromatography, and a highly pure preparation obtained through chromatofocusing chromatography (Manthey et al., 1990). Isoelectric focusing was carried out in commercially available LKB PAG plates, pH 3.5-9.5. Sections of gels to be stained for aldehyde dehydrogenase activity were immersed in 32 mM tetrasodium pyrophosphate buffer, pH 8.2, containing 1 mM pyrazole, 2 mM NAD, 0.8 mg/ml nitroblue trazolium, 0.05 mg/ml phenazine methosulfate and 5 mM acetaldehyde. Control gels lacked either tissue fraction, NAD or acetaldehyde.

Fig. 2.

Zymogram of the cytosolic fraction from 191 embryonic retinas, semi-pure (AHD-2) and highly purified AHD-2 separated by isoelectric focusing. Up is towards basic and down towards acidic pH. Purified AHD-2 runs as four closely spaced bands, probably corresponding to four conformational states, with pl values of 8.16, 8.09, 8.03 and 7.88 as marked by triangles. These signals line up precisely with four signals in the retina sample; in contrast to the pure preparation, however, the retinal signal at pH 8.16 exceeds the other three in intensity. In addition to AHD-2 several other bands indicative of dehydrogenases are visible here, except for the band at pH 6.23 which was due to non-specific dye reaction.

Fig. 2.

Zymogram of the cytosolic fraction from 191 embryonic retinas, semi-pure (AHD-2) and highly purified AHD-2 separated by isoelectric focusing. Up is towards basic and down towards acidic pH. Purified AHD-2 runs as four closely spaced bands, probably corresponding to four conformational states, with pl values of 8.16, 8.09, 8.03 and 7.88 as marked by triangles. These signals line up precisely with four signals in the retina sample; in contrast to the pure preparation, however, the retinal signal at pH 8.16 exceeds the other three in intensity. In addition to AHD-2 several other bands indicative of dehydrogenases are visible here, except for the band at pH 6.23 which was due to non-specific dye reaction.

Fig. 3.

HPLC/spectrophotometric assays for NAD-dependent retinoic acid synthesis. The upper row shows synthesis in whole embryonic retinas: 8 E14-15 retinas per run synthesized 1.40 nmole of retinoic acid, corresponding to a rate of 36.0 pmole/minute/retina; this converted 5.6% of retinaldehyde to retinoic acid. The negative controls either lacked tissue, or NAD or retinaldehyde. Retention time for retinoic acid was ∼6.0 minutes, and retention time for retinaldehyde 13.3 minutes. The lower row gives a comparison of 58 dorsal and ventral retina halves from E13-14.5. The dorsal retinas produced 14.05 nmole of retinoic acid, corresponding to a rate of 24.2 pmole/minute/retina; and the ventral retina 1.52 nmole, corresponding to a rate of 2.6 pmole/minute/retina. In dorsal retinas 56.2% of retinoic acid was converted to retinaldehyde as compared to 6.1% in ventral retinas, indicating that the true dorsal rate was considerably higher than 24.2.

Fig. 3.

HPLC/spectrophotometric assays for NAD-dependent retinoic acid synthesis. The upper row shows synthesis in whole embryonic retinas: 8 E14-15 retinas per run synthesized 1.40 nmole of retinoic acid, corresponding to a rate of 36.0 pmole/minute/retina; this converted 5.6% of retinaldehyde to retinoic acid. The negative controls either lacked tissue, or NAD or retinaldehyde. Retention time for retinoic acid was ∼6.0 minutes, and retention time for retinaldehyde 13.3 minutes. The lower row gives a comparison of 58 dorsal and ventral retina halves from E13-14.5. The dorsal retinas produced 14.05 nmole of retinoic acid, corresponding to a rate of 24.2 pmole/minute/retina; and the ventral retina 1.52 nmole, corresponding to a rate of 2.6 pmole/minute/retina. In dorsal retinas 56.2% of retinoic acid was converted to retinaldehyde as compared to 6.1% in ventral retinas, indicating that the true dorsal rate was considerably higher than 24.2.

Fig. 4.

F9 reporter cells incubated overnight with pieces of dorsal (D) and ventral (V) retina and cortex (Cx) from E13.5 mouse, and reacted with X-Gal Some of the tissue pieces became dislodged or were washed away during processing. Scale 1 mm.

Fig. 4.

F9 reporter cells incubated overnight with pieces of dorsal (D) and ventral (V) retina and cortex (Cx) from E13.5 mouse, and reacted with X-Gal Some of the tissue pieces became dislodged or were washed away during processing. Scale 1 mm.

For the high-pressure liquid chromatography (HPLC) assay, the tissue was homogenized in 20 mM TEA-HC1, pH 7.4, in the presence of 1 mM DTT and 0.1 mM EDTA, and spun for 1 hour at 105,000 g in a Beckman Airfuge. Enzyme activity was determined at 37°C, with the final reaction mixture containing 2.5% dimethylsulfoxide, 4 mM NAD, 32 mM tetrasodium pyrophosphate pH 8.2, 0.1 mM pyrazole, 5 mM glutathione, 1 mM EDTA, 25 μM all-trans-retinalde-hyde, and cytosol from homogenized tissue. After 5 or 10 minutes of incubation the reaction was stopped by immersion in ice-water and the samples were processed for HPLC/ spectrophotometric assay as described (Lee et al., 1991).

The production of the F9 retinoic acid reporter cells is being described by Wagner and Jessell (in preparation). The cells were maintained in L15 CO2 tissue culture medium (Specialty Media) with 20% fetal calf serum in gelatin-coated tissue culture flasks. For assays of tissue fragments, they were transferred to gelatin-coated coverslips mounted in 35 × 10 mm Petri dishes, and for enzyme assays they were grown in 96-well tissue culture plates.

For the semi-quantitative assays, the tissues were homogenized in 60 – 90 of complete L15 medium at 4°C, followed by freeze-thawing and separation at 4°C into either cytosol (105,000 g for 1 hour), or cytosol with microsomes fractions (12,000 g for 20 seconds), or complete homogenate without debris (1 g for 1 hour supernatant). In all experiments in which the tissue was homogenized in tissue culture medium, all detectable retinoic acid activity and synthesis were confined to the high-speed supernatant (but see legend to Fig. 9). The estimates of comparative tissue levels (Fig. 5) were done with complete homogenates of roughly similar-sized tissue chunks, directly pipetted onto the reporter cells in a 96-well plate and incubated at 37°C overnight in a tissue culture incubator. For the synthesis experiments in Fig. 6, between 14 and 22 dorsal and ventral retina halves were preincubated for 3 hours at 37°C with 5×lQ∼7 M retinaldehyde or retinol or β-carotene, in combination with 1.2 mM NAD. The samples were plated at serial dilutions onto the reporter cells and incubated overnight.

Fig. 5.

(left) Low-power view of a frontal section through an E10.5 mouse labeled with aldehyde dehydrogenase antisera (Lindahl and Evces, 1984a,b; Russo and Hilton, 1988). The section passed through the brain at the level of the diencephalon (de) and mesencephalon (me), which are negative for the antisera. On the right side in the micrograph the eye cup with invaginating lens placode (ey) is brightly labeled, and on the left side, which is slightly further caudal, only the labeled surface ectoderm of the eye region is visible. The maxiliar process (mp) and mandibular arch (ma) are negative. In the heart region (li) aldehyde dehydrogenase immunoreactivity is just appearing in the pericardial sac. The hindlimb buds and the spinal cord (sc) are negative. Scale 500 μm. (Right) Estimates of comparative retinoic acid levels in different organs of an Ell mouse, measured in tissue homogenates with the F9 reporter cells.

Fig. 5.

(left) Low-power view of a frontal section through an E10.5 mouse labeled with aldehyde dehydrogenase antisera (Lindahl and Evces, 1984a,b; Russo and Hilton, 1988). The section passed through the brain at the level of the diencephalon (de) and mesencephalon (me), which are negative for the antisera. On the right side in the micrograph the eye cup with invaginating lens placode (ey) is brightly labeled, and on the left side, which is slightly further caudal, only the labeled surface ectoderm of the eye region is visible. The maxiliar process (mp) and mandibular arch (ma) are negative. In the heart region (li) aldehyde dehydrogenase immunoreactivity is just appearing in the pericardial sac. The hindlimb buds and the spinal cord (sc) are negative. Scale 500 μm. (Right) Estimates of comparative retinoic acid levels in different organs of an Ell mouse, measured in tissue homogenates with the F9 reporter cells.

Fig. 6.

Retinoic acid synthesis by cytosolic fractions from dorsal and ventral embryonic retina halves from retinol (left), βcarotene (middle) and retinaldehyde (right).Fig. 7. (A) Retinoic acid synthesis by cytosolic fractions from dorsal and ventral embryonic retina halves preincubated with 5×10-7 and 2.5×10-5 M retinaldehyde and 1.2 mM NAD, and measured as dilutions onto the reporter cells. (B) Ratios between dorsal and ventral retinoic acid synthesis from retinaldehyde, as a function of concentration of substrate.

Fig. 6.

Retinoic acid synthesis by cytosolic fractions from dorsal and ventral embryonic retina halves from retinol (left), βcarotene (middle) and retinaldehyde (right).Fig. 7. (A) Retinoic acid synthesis by cytosolic fractions from dorsal and ventral embryonic retina halves preincubated with 5×10-7 and 2.5×10-5 M retinaldehyde and 1.2 mM NAD, and measured as dilutions onto the reporter cells. (B) Ratios between dorsal and ventral retinoic acid synthesis from retinaldehyde, as a function of concentration of substrate.

Fig. 7.

(A) Retinoic acid synthesis by cytosolic fractions from dorsal and ventral embryonic retina halves preincubated with 5×10-7 and 2.5×10-5 M retinaldehyde and 1.2 mM NAD, and measured as dilutions onto the reporter cells. (B) Ratios between dorsal and ventral retinoic acid synthesis from retinaldehyde, as a function of concentration of substrate.

Fig. 7.

(A) Retinoic acid synthesis by cytosolic fractions from dorsal and ventral embryonic retina halves preincubated with 5×10-7 and 2.5×10-5 M retinaldehyde and 1.2 mM NAD, and measured as dilutions onto the reporter cells. (B) Ratios between dorsal and ventral retinoic acid synthesis from retinaldehyde, as a function of concentration of substrate.

In the enzyme inhibitor assays, the inhibitors were allowed to act on the homogenates for 10–20 minutes at 4°C; then retinaldehyde and NAD were added, and the samples were processed as described above. After an exposure time of about 12 hours, the reporter cells were fixed with 1%glutaraldehyde, and β-galactosidase activity was visualized with 5-bromo-4-chloro-3-indolyl-β-D-galactopyranoside (X-Gal). The intensity of the blue product was measured with a microtiter plate reader (ELISA reader, Fisher Biotech BT100) at 405 nm. Only the comparative values of colorimetric readings within one assay were of significance. The absolute values varied between experiments, because the reporter cells tended to become less responsive with successive passages perhaps due to gradual loss of the plasmid.

For the zymography-bioassay (Fig. 9) the tissue was homogenized in 70 μ1 buffer containing 10 mM sodiumphosphate, 30 mM NaCI, 0.2% sodium-azide, 10 mg/ml aprotinin, 1 mM PMSF, 0.2 mM EDTA and 0.1% 2-mercaptoethanol, and spun for 20 seconds at 12,000 g. Isoelectric focusing was carried out in Isogel agarose IEF plates (FMC BioProducts) according to the manufacturer’s instructions. Gel lanes were sectioned into parallel series of slices with a cutting device that spanned four neighboring lanes and consisted of 40 blades spaced —1.8 mm apart. The slices were transferred rapidly into 96-well plates containing 100 μ1 of L15 medium per well. Then 10-7 M retinaldehyde, 1.2 mM NAD and 2 mM DTT were added, and the plates were incubated for synthesis at 37°C. After 3 hours the media samples were transferred into 96-well plates with confluent cultures of reporter cells, and processed as described above.

The cytosolic fraction of 191 retinas from E13-14 mice was isolated and separated by isoelectric focusing in parallel with a highly pure and a semi-pure preparation of AHD-2 from adult liver (Manthey et al., 1990). Enzyme activity was visualized zymographically by NAD-dependent oxidation of acetaldehyde. The highly purified enzyme appeared as four closely spaced bands in the basic portion of the gel between pH 7.9-8.2, which lined up exactly with basic signals in the retina sample (triangles in Fig. 2). This match identified the retinal form unambiguously as AHD-2. In previous immunoblots with the same aldehyde dehydrogenase antisera as shown in Fig. 1 (Lindahl and Evces, 1984a,b; McCaffery et al., 1991; Russo and Hilton, 1988), the basic group of bands was the only signal detectable in the early embryonic retina. In the zymogram of Fig. 2, several additional bands indicated NAD-dependent acetaldehyde oxidation by enzymes not recognized by the antisera; their identity was not further examined.

Cytosolic fractions of embryonic retinas were tested for retinoic acid synthesis from retinaldehyde by a spectrophotometrical high-pressure liquid chromatography (HPLC) detection system. Fig. 3, upper row, illustrates the NAD-dependent retinoic acid synthesis in whole retinas from E14-15 mouse embryos. In Fig. 3, lower row, the dorsal and ventral halves from 58 E13-14 retinas are compared: the area of the retinoic acid peak in the dorsal trace was ten times greater than the ventral peak. Although such a dorsoventral ratio in retinoic acid synthesis seemed consistent with the difference in AHD-2 levels, the result might not reflect the in vivo situation, because only the cytosolic fractions were tested and, in particular, because the spectrophotometrical detection system required high concentrations of substrate: here 25 μM retinaldehyde was used.

For the estimation of retinoic acid synthesis in vivo we used a reporter cell line derived by transfection of F9 teratocarcinoma cells with a plasmid containing the β galactosidase gene under control of the retinoic acid responsive element from the retinoic acid receptorβ (Wagner and Jessell, in preparation). Freshly dissected pieces from dorsal and ventral E13.5 retinas and cerebral cortex were grown overnight on top of a confluent layer of F9 cells, and the induced β-galactosidase synthesis was visualized with the X-Gal color reagent (Fig. 4). The retinas induced much more reaction product than the cortex (Cx). In contrast to the HPLC results, however, the ventral retina (V) was significantly more active than the dorsal retina (D).

Higher retinoic acid synthesis in ventral as compared to dorsal retina was consistently seen throughout embryonic development and in the early postnatal mouse, and a similar dorsoventral ratio was observed in embryonic rat and chick retinas. No asymmetry was seen in the anteroposterior axis.

As the color reaction of F9 cells incubated with tissue fragments did not easily lend itself to quantification, freshly dissected tissues were homogenized in the F9 culture medium, the homogenate was placed overnight onto F9 cells growing in 96-well plates, and the X-Gal reaction was evaluated colorimetrically. The color readings with this semiquantitative method were similar to the visual comparisons with the explant method: the ventral retina gave about twice the dorsal readings and the cortex was very low. Titration curves run in parallel with commercial retinoic acid were roughly linear in their mid-range, i.e. doubling in colorimetric readings corresponded to a doubling in applied retinoic acid. A comparative screen of different tissues of the E11.0 mouse with this method is shown in Fig. 5, right’, at this age the spinal cord and the retinas exceeded all other tissues in retinoic acid levels. A low-power view of a frontal section through an E10.5 mouse labeled with the aldehyde dehydrogenase antiserum is shown in the micrograph of Fig. 5: the retina and the overlying surface ectoderm are brightly labeled here, but no immunoreactivity is detectable in the spinal cord. Thus both the overall retinoic acid estimates, as well as the dorsoventral retina comparisons, indicated that AHD-2 can only account for a fraction of the total synthesis.

The color reactions obtained with the tissue homogenates seemed to reflect not only the retinoic acid levels at the time of homogenization but also synthesis during the incubation. This second aspect was exploited for the analysis of the problem of retinoic acid asymmetry in the dorsoventral axis, as the method allowed testing of isolated subcellular fractions, different substrates, coenzymes and inhibitors. The main focus was on the evaluation of differences between dorsal and ventral retina halves, as all artefacts should affect both samples in the same way.

When NAD was added to homogenates of dorsal and ventral retina halves (not shown), retinoic acid synthesis in both preparations was increased to a similar extent: 1.2 mM NAD gave an approximate quadrupling in colorimetric readings, corresponding to a fourfold increase in retinoic acid, as estimated from retinoic acid standard curves run on the same plate. NADH could not be evaluated, as it interacted directly with the F9 cells to cause an increase in β-galactosidase synthesis; the mechanism of this effect is not known. The amount of NADH produced during the enzymatic reactions, however, was not large enough to make a noticeable contribution to the color reaction, as tested by NAD-dependent oxidation of propionaldehyde (propionaldehyde was oxidized by the same enzymes as retinaldehyde, because it acted as a competitive inhibitor of retinoic acid synthesis in the embryonic retina; see Table 1). NADP had no significant effect.

Table 1.

Effects of enzyme inhibitors on retinoic acid synthesis in dorsal (D) and ventral (V) retinas

Effects of enzyme inhibitors on retinoic acid synthesis in dorsal (D) and ventral (V) retinas
Effects of enzyme inhibitors on retinoic acid synthesis in dorsal (D) and ventral (V) retinas

Retinoic acid synthesis was further enhanced by addition of retinol in combination with NAD. Under the conditions illustrated in Fig. 6, left, the dorsoventral ratio in colorimetric readings, which in the absence of additives was about 1:2, approached 1:1. All measurements were done as serial dilutions of preincubated samples in order to assure that they included the linear range of the reporter cell responses, because at low levels the readings were less accurate and high concentrations of retinoids are toxic to F9 cells (Strickland and Mahdavi, 1978). In most experiments, the tissue homogenates were preincubated with the additives at 37°C for 3 hours; shorter incubation times were generally less effective but gave similar dorsoventral ratios in all cases, excluding significant differences in reaction rates. In addition, the additives were diluted out in the absence of tissue to test for direct effects on the F9 cells.

Beta-carotene could also serve as a precursor for retinoic acid synthesis in the embryonic retina, but it probably plays a minor role, as judged by comparison with adult liver which converted β-carotene much more effectively into retinoic acid (not shown). The ventral colorimetric readings consistently exceeded the dorsal readings (Fig. 6, middle), which is an additional indication for a minor role of this pathway, as concluded from the observations reported below.

Our attention focused on the retinaldehyde-to-retinoic acid conversion, the step common to both retinol and β-carotene usage, as the spatial distribution of AHD-2 pointed to a dorsoventral asymmetry at this level. Both dorsal and ventral retinas were able to metabolize retinaldehyde in combination with NAD. At the highest retinaldehyde concentrations, the dorsal titration curves often peaked and fell (Fig. 6, right). This phenomenon was more pronounced with higher tissue amounts and longer preincubation times, and it was probably due to toxic effects of retinoic acid or a related retinoid on the reporter cells (Strickland and Mahdavi, 1978).

At retinaldehyde concentrations of 5x10-7 M the dorsal synthesis exceeded ventral synthesis by a small fraction. This difference was substantially more marked at 2.5×10-5 M retinaldehyde (Fig. 7A). Fig. 7B shows the dorsoventral ratios in colorimetric readings as a function of the retinaldehyde levels. These ratios indicate that the dorsal pathway is more effective in handling high substrate levels than the ventral pathway, thus the F9 bioassay supported the HPLC results (Fig. 3) done at high substrate concentrations. In both dorsal and ventral retina, an absolute requirement for NAD was found, indicating dehydrogenases were solely responsible for retinoic acid synthesis.

We tested enzyme inhibitors commonly used in the analysis of the retinoic acid synthetic pathway. Several inhibitors, such as chloral hydrate, cyanide and metal chelating reagents, could not be tested because of their toxicity to the F9 cells, and for the same reason we were limited to non-toxic dosages of other inhibitors. A summary of the most significant inhibitor effects is given in Fig. 8 and a complete list of the inhibitors tested in Table 1. We did not find any indication for a dorsoventral difference in retinoic acid turnover, as ketoconazole, an inhibitor of P450-mediated retinoic acid breakdown (Napoli, 1986), did not have a noticeable effect on overall retinoic acid levels or on the dorsoventral ratio. Disulfiram, an inhibitor of cytosolic aldehyde dehydrogenase (Greenfield and Pietruszko, 1977; Vallari and Pietruszko, 1982), depressed dorsal synthesis at very low concentrations (6 ×M) while having little effect on the ventral pathway. At 50 ×M disulfiram both pathways were strongly inhibited.

Fig. 8.

Summary diagram of inhibitor effects on retinoic acid synthesis from retinaldehyde in dorsal and ventral embryonic retina halves, taken from two experiments.

Fig. 8.

Summary diagram of inhibitor effects on retinoic acid synthesis from retinaldehyde in dorsal and ventral embryonic retina halves, taken from two experiments.

Fig. 9.

Zymography-bioassay of 24 dorsal (D) and ventral (V) halves of E12.5 retinas. The positions of focused pH markers (Sigma) are indicated at the bottom. AHD-2 activity, focusing in the basic portion of the gel, is practically restricted to the dorsal trace. The activity peak at pH 5.7 was consistently substantially higher in ventral than dorsal samples. The position of the practically ventral-specific activity in the acidic portion of the gel does not reflect its charge, but the origin of the sample. This activity behaved as only partly cytosolic: while it appeared to be completely soluble under the high-salt homogenization conditions in L15 medium, under the lower-salt conditions in the isoelectric focusing protocol it was partly present in the high-speed pellet, but was brought into solution by washing of the pellet. A similar erratic behavior in isoelectric focusing gels has previously been observed for the cytosolic isoform AHD-11, which however is not thought to convert retinaldehyde to retinoic acid (Manthey et al., 1990).

Fig. 9.

Zymography-bioassay of 24 dorsal (D) and ventral (V) halves of E12.5 retinas. The positions of focused pH markers (Sigma) are indicated at the bottom. AHD-2 activity, focusing in the basic portion of the gel, is practically restricted to the dorsal trace. The activity peak at pH 5.7 was consistently substantially higher in ventral than dorsal samples. The position of the practically ventral-specific activity in the acidic portion of the gel does not reflect its charge, but the origin of the sample. This activity behaved as only partly cytosolic: while it appeared to be completely soluble under the high-salt homogenization conditions in L15 medium, under the lower-salt conditions in the isoelectric focusing protocol it was partly present in the high-speed pellet, but was brought into solution by washing of the pellet. A similar erratic behavior in isoelectric focusing gels has previously been observed for the cytosolic isoform AHD-11, which however is not thought to convert retinaldehyde to retinoic acid (Manthey et al., 1990).

The reverse effect was seen with p-hydroxymercuri-benzoate (pHMB), an inhibitor of sulfhydryl-dependent enzymes, which strongly depressed mainly the ventral pathway at 40 ×M concentration and inhibited both pathways at 80 ×M. Quinacrine, an inhibitor of complex properties most commonly used as indicator of flavoenzymes (Webb, 1966), exhibited a biphasic effect depending on its concentration: at low levels it depressed synthesis in both dorsal and ventral samples, and at higher concentrations it enhanced dorsal synthesis while still depressing ventral synthesis; addition of flavin adenine dinucleotide (FAD) did not reverse the quinacrine effect. Allopurinol, an inhibitor of xanthine oxidase, failed to show an effect, as did two inhibitors of aldehyde oxidase, propylgallate and acridinylamino-methanesulfon-m-aniside (mAMSA) (Gormley et al., 1983). Propionaldehyde and citral (Connor and Smit, 1987) inhibited both pathways competitively with insignificant differences.

The comparisons of retinoic acid synthesis between dorsal and ventral retina pointed to the existence of dehydrogenase(s) in ventral retina different from the AHD-2 isoform in dorsal retina. Our attempts to visualize directly the two pathways by zymography, the commonly used method for the visualization of dehydrogenases (see Fig. 2), were only partially successful: AHD-2 was obvious in all dorsal samples tested, but we saw no convincing evidence for a ventral-specific pathway. The problem seemed to lie in the method: detection of enzymatic activity by conventional zymo-graphy is rather insensitive for embryonic applications and, more importantly, it does not work with retinaldehyde as substrate, but it is usually done with lower aldehydes such as acetaldehyde.

For a more sensitive and specific analysis, we used the reporter cells in a zymography-bioassay. Parallel lanes of isoelectric focusing gels containing separated proteins from dorsal and ventral retinas were cut into thin slices, from each of which the proteins were extracted into culture medium and assayed for retinoic acid synthesizing capacity. This assay required only small amounts of tissue, and it very reliably demonstrated not only AHD-2 in dorsal samples, but two enzymatic activities accounting for synthesis in ventral retina (Fig. 9). One of these, which focused around pH 5.6-5.7, was also present in the dorsal samples but at lower amounts. The other was practically ventral-specific; its pl could not be determined, because its position in the gel depended partly on the position of the origin of the sample, which either points to a loose membrane association or to the formation of aggregates (see legend to Fig. 9).

During the first two postnatal weeks, the relative maximum of retinoic acid content and synthetic capacity shifted gradually from ventral to dorsal retina. The details of this shift, which reflects a slow disappearance of the ventral enzymes, will be described elsewhere. In the adult retina the anatomical AHD-2 distribution is largely restricted to the dorsal third (Fig. 1) (McCaffery et al., 1991), and when adult retinas were dissected into dorsal, medial and ventral fragments for the explant assays on F9 cells, the most intense color reaction was obtained with the dorsal third (Fig. 10, left). In tests of retinoic acid synthesis a similar distribution was apparent, and the dorsal third was most susceptible to disulfiram inhibition (Fig. 10, right).

Fig. 10.

(Left) F9 reporter cells incubated overnight with pieces of dorsal (D), medial (M) and ventral (V) retina from adult mice. Scale 1mm. (Right) Diagram of retinoic acid synthesis from retinaldehyde by dorsal, medial and ventral thirds from adult retina, and inhibition by 6 μM disulfiram.

Fig. 10.

(Left) F9 reporter cells incubated overnight with pieces of dorsal (D), medial (M) and ventral (V) retina from adult mice. Scale 1mm. (Right) Diagram of retinoic acid synthesis from retinaldehyde by dorsal, medial and ventral thirds from adult retina, and inhibition by 6 μM disulfiram.

A major organizational principle of the brain is the ordered representation of neuronal information in the form of multiple and interconnected topographical maps. This pattern has been most intensively studied on the projections between retinas and optic tectum, a component of the central nervous system more easily accessible to experimental manipulations than the rest of the brain. The basic layout of retinotectal projection maps is identical throughout all vertebrates. The mechanism of map formation is not understood, but it is believed to reflect biochemical differences in the dorsoventral and anteroposterior axes of the retina which are set up very early in development (Bonhoeffer and Gierer, 1984; Sperry, 1963). We previously described a biochemical difference for the dorsoventral axis: a cytosolic aldehyde dehydrogenase present at high concentrations in the dorsal embryonic retina (McCaffery et al., 1991), and here we identified it as the murine isoform AHD-2, which corresponds to the human isoform El (Greenfield and Pietruszko, 1977; Manthey et al., 1990). The most widely studied aspect of aldehyde dehydrogenases, the detoxification of exogenous substances, did not seem relevant to the localization in the embryonic eye; rather the striking anatomical distribution pointed to a developmental role of an endogenous substrate or product of aldehyde dehydrogenase. Here we pursued the lead given by the observation that AHD-2 is the major enzyme in the liver of the adult mouse capable of NAD-dependent retinoic acid synthesis (Lee et al., 1991).

Although retinoic acid has not directly been implicated in retinal development, several observations are consistent with such a role. Both too little and too much of vitamin A or retinoic acid during embryogenesis cause severe eye defects (Geelen, 1979; Giroud and Martinet, 1961; Giroud et al., 1962; Hale, 1937; Lammer et al., 1985; Warkany and Schraffenberger, 1946; Wilson et al., 1953), and the embryonic eye contains protein components involved in retinoic acid action: (1) cellular retinol binding protein (CRBP), which is thought to facilitate the local accumulation of the precursor retinol (Dollé et al., 1990); (2) cellular retinoic acid binding protein (CRABP), which may act as a retinoic acid buffer (Hirschel-Scholz et al., 1989; Vaessen et al., 1990); and (3) the retinoic-acid receptor α(RARα) (Dollé et al., 1990). CRBP and R A Rα-are uniformly distributed in the early embryonic retina (Dollé et al., 1990), and CRABP is first expressed in central retina in the early differentiating ganglion cells (Dollé et al., 1990; Perez-Castro et al., 1989).

Retinoic acid can be generated from two dietary precursors: retinol, an animal product, and β-carotene, a plant product. In a reversible reaction retinol is oxidized by alcohol dehydrogenase to retinal (=retinal-dehyde), which undergoes irreversible oxidation to retinoic acid. Beta-carotene can either be cleaved centrally into two molecules of retinal, or it is broken down by eccentric cleavage to apocarotenals, which are thought to undergo β-oxidation to retinal. In addition, several observations point to the existence of a conversion of β-carotene to retinoic acid that does not proceed via retinal (Napoli and Race, 1988; Wang et al., 1991).

The conversion of retinal to retinoic acid is thought to be catalyzed by three types of enzymes: aldehyde dehydrogenase, aldehyde oxidase and xanthine oxidase (Bhat et al., 1988; Futterman, 1962; Lakshmanan et al., 1964; Lee et al., 1991; Leo et al., 1987; Napoli, 1986; Napoli and Race, 1988; Siegenthaler et al., 1990; Williams and Napoli, 1985). Of these only the aldehyde dehydrogenases have been tested as purified preparations (Ambroziak and Pietruszko, 1991; Lee et al., 1991). Actions of the other enzymes have been determined in whole cells or in tissue extracts and their implication is indirect, inferred from reaction characteristics and inhibitor profiles; it is not clear how many enzymes are involved. Different studies report very different reaction characteristics, which could reflect variable contributions of the mentioned enzymes or other as yet unidentified enzymes. Moreover, these types of enzymes are not specific to one particular substrate, rather they all act on a range of substrates but with different enzymes showing different ranges of preference. In general, questions of retinoic acid synthesis are considered as too complex and are usually not discussed in the context of morphogenetic mechanisms (Brockes, 1989; Tabin, 1991). The present observations, however, indicate that at least for the retina the topic cannot be neglected, as the morphogenetically significant feature here appears to be the spatial arrangement of synthetic pathways rather than receptors and binding proteins.

A major problem for studies of retinoic acid synthesis in the embryo are technical limitations. The most commonly used and most sensitive method is HPLC, which is nevertheless rather cumbersome, requiring tissue amounts that are exorbitant on an embryonic scale. Here we applied a standard HPLC protocol to the question of dorsoventral asymmetry, and the result seemed to confirm the prediction from the AHD-2 localization: a much higher retinoic acid synthesis in the dorsal retina. Attempts to verify this observation on intact tissue with help of an F9 reporter cell line, however, gave the diametrically opposite result: higher retinoic acid synthesis ventrally than dorsally. In order to solve the dilemma, we turned the F9 tests into semiquantifiable enzymatic microassays on a scale compatible with embryonic tissue amounts. Because performing the enzymatic assays in tissue culture medium added much complexity, we focused on comparative observations. Although the tissue homogenates in the absence of additives clearly represented an artificial situation, they gave similar color reactions to the explants of intact tissue pieces, which are probably rather close to the in vivo condition.

The enzymatic tests showed that a key difference between dorsal and ventral retinoic acid synthesis exists in the final step of retinaldehyde oxidation. The dorsal pathway functioned effectively at high retinaldehyde levels, at which the ventral pathway became saturated or substrate-inhibited. The ventral pathway was relatively more effective at low and probably physiological retinaldehyde levels. The pronounced sensitivity of the dorsal pathway to disulfiram, which at low concentrations is a very specific inhibitor of the El/AHD-2 aldehyde dehydrogenase (Vallari and Pietruszko, 1982), and the zymography-bioassay (Fig. 9) provided compelling evidence for AHD-2 as the main source of retinoic acid in dorsal retina.

Ventral synthesis was found to be mediated by enzymatic activities different in charge from AHD-2 (Fig. 9). Their NAD dependence unambiguously defined them as dehydrogenases, although the two inhibitors that affected them preferentially, pHMB and quinacrine, have been previously used as arguments in favor of aldehyde and xanthine oxidases in retinoic acid synthesis (Bhat et al., 1988; Futterman, 1962; Lakshmanan et al., 1964). As both of these inhibitors lack specificity towards particular enzymes (Webb, 1966), however, we take their effects only as additional confirmation for the different enzymatic pathways in dorsal and ventral retina. So far, we do not know the precise identity of the ventral enzymatic activities, as they do not resemble any of the retinoic acid synthesizing enzymes characterized in the fiver of the adult mouse (Lee et al., 1991). They might represent enzymes not previously implicated in retinoic acid synthesis at all, or enzymes that remain to be identified in the mouse, such as a murine homolog to the human cytosolic aldehyde dehydrogenase E3 (Abe et al., 1990; Kurys et al., 1989). E3 is distinguished from El by lower disulfiram sensitivity and very effective conversion of γ-aminobutyraldehyde to γ-aminobutyric acid (GABA), the alternate synthetic pathway of GABA from putrescine; we found a substantially higher GABA content in ventral than dorsal embryonic retina (Eliasson, Baughman and Drager, unpublished observation). Like El, E3 catalyses the conversion of retinal to retinoic acid (Ambroziak and Pietruszko, 1991).

Our crude assays of retinoic acid distribution in the embryo (see Fig. 5) placed the retinas among the richest organs after the spinal cord, and most of the lower but still substantial levels in dorsal retina were generated by AHD-2. Nevertheless, the performance of AHD-2 is, in fact, very poor in view of its tissue concentrations. In the dorsal retina of the younger embryo AHD-2 makes up about 1% of soluble protein, a level that far exceeds normal enzyme levels and is more comparable to levels of structural proteins. Even its much lower levels in ventral retina are still high for an enzyme (McCaffery et al., 1991). By comparison, the ventral enzymes, which are at least twice as efficient in retinoic acid synthesis (Fig. 9), are contained in such low-protein fractions that we have so far not been able to identify them unambiguously in protein gels (preliminary observations). The unusual levels of AHD-2 raise the question of an additional, possibly non-enzymatic function. There is a precedent for such a suggestion: an androgen-binding protein, which is normally abundant in certain fibroblasts but missing in patients with testicular feminization, was recently identified as enzymatically active El/AHD-2 (Pereira et al., 1991).

The high levels of retinoic acid in the embryonic retina ought to stimulate transcription of the retinoic acid receptor β (RARβ), which is inducible through a very sensitive retinoic acid responsive element (RARE) in its promotor (the system used in the reporter cells of Wagner and Jessell) (de Thé et al., 1990; Sucov et al., 1990). However, the normal embryonic retina contains only RARα and no RARβ transcripts (Dollé et al., 1990). Mice transgenic for βgalactosidase under con-trol of a short 34-bp stretch containing the RARE of the RARβ promotor, show strong expression of the reporter gene in the eye region (Rossant et al., 1991), but similar transgenic mice in which β-galactosidase is under control of a long stretch, containing 3.8 kb of the 5’-flanking sequence of the RARβ gene, do not express the transgene in the neural retina (Mendelsohn et al., 1991). The lacZ staining pattern in the mice carrying the minimal promotor construct appears to reflect tissue levels of retinoic acid, as judged from the uniform inducibility of the transcript by teratogenic doses of retinoic acid (Rossant et al., 1991) and a comparison with our retinoic acid estimates in dissected embryos (Fig. 5). The long control sequence, however, evidently contains elements directing tissue-specific expression of the RARβ, including a mechanism for the strong repression in the neural retina which overrides the induction by retinoic acid (Mendelsohn et al., 1991).

The retinoic acid system we described here for the embryonic mouse retina appears to be universal in vertebrates, as it applies also to embryonic rat and chick: their retinas contained overall high levels of retinoic acid; ventral levels exceeded dorsal levels; synthesis in dorsal and ventral samples was mediated by different enzymes; and the dorsal enzyme resembled AHD-2 including sensitivity to disulfiram (unpublished observations). Because of this high degree of conservation, and because of the early appearance of retinoic acid in the embryonic eye, the system is a good candidate for being an essential agent in the specification of the dorsoventral axis. It may influence the expression of other components asymmetrically distributed in the dorsoventral axis, such as the transcript Pax2 in ventral retina (Nomes et al., 1990) and the accessibility of the protein-translation factor p40 in dorsal retina (McCaffery et al., 1990; Rabacchi et al., 1990). Eventually it may direct the cell-surface expression of factors that could play a role in the growthcone affinity for central targets (Constantine-Paton et al., 1986; Trisler et al., 1981).

We are grateful to Dr. Ronald Lindahl and Dr. John Hilton for gifts of the aldehyde dehydrogenase antisera. Acridinyla-mino-methanesulfon-m-aniside (mAMSA) was donated by the Drug Synthesis and Chemistry Branch of the National Cancer Institute. The work was supported by NIH grants EY 03819 (U.C.D.) and CA 21737 (N.E.S.), and by Howard Hughes Medical Institute Funds (M.A.W.).

Abe
,
T.
,
Takada
,
K.
,
Ohkawa
,
K.
and
Matsuda
,
M.
(
1990
).
Purification and characterization of a rat brain aldehyde dehydrogenase able to metabolize y-aminobutyraldehyde to γ-aminobutyric acid
.
Biochem. J
.
269
,
25
29
.
Ambroziak
,
W.
and
Pietruszko
,
R.
(
1991
).
Human aldehyde dehydrogenase. Activity with aldehyde metabolites of monoamines, diamines, and polyamines
.
J. Biol. Chem
.
266
,
13011
13018
.
Benbrook
,
D.
,
Leinhardt
,
E.
and
Pfahl
,
M.
(
1988
).
A new retinoic acid receptor identified from a hepatocellular carcinoma
.
Nature
333
,
669
672
.
Bhat
,
P. V.
,
Poissant
,
L.
and
Lacroix
,
A.
(
1988
).
Properties of retinal-Retinoic acid synthesis in the retina 381 oxidizing enzyme activity in rat kidney
.
Biochim. Biophys. Acta
967
,
211
217
.
Bonhoeffer
,
F.
and
Gierer
,
A.
(
1984
).
How do retinal axons find their targets on the tectum?
TINS
7
,
378
381
.
Brand
,
N.
,
Petkovich
,
M.
,
Krust
,
A.
,
Chambón
,
P.
,
De Thé
,
H.
,
Marchio
,
A.
,
Tiollais
,
P.
and
Dejean
,
A.
(
1988
).
Identification of a second human retinoic acid receptor
.
Nature
332
,
850
853
.
Brockes
,
J. P.
(
1989
).
Retinoids, homeobox genes, and limb morphogenesis
.
Neuron
2
,
1285
294
.
Connor
,
M. J.
and
Smit
,
M. H.
(
1987
).
Terminal-group oxidation of retinol by mouse epidermis. Inhibition in vitro and in vivo
.
Biochem. J
.
244
,
489
492
.
Constantine-Paton
,
M.
,
Blum
,
A. S.
,
Mendez-Otero
,
R.
and
Barnstable
,
C. J.
(
1986
).
A cell surface molecule distributed in a dorso-ventral gradient in the perinatal rat retina
.
Nature
324
,
459462
.
de Thé
,
H.
,
del Mar Vivanco-Rulz
,
M.
,
Tiollais
,
P.
,
Stunnenberg
,
H.
and
Dejean
,
A.
(
1990
).
Identification of a retinoic acid responsive element in the retinoic acid receptor ft gene
.
Nature
343
,
177
180
.
Dollé
,
P.
,
Ruberte
,
E.
,
Kastner
,
P.
,
Petkovich
,
M.
,
Toner
,
C. M.
,
Gudas
,
L.
and
Chambón
,
P.
(
1989
).
Differential expression of the genes encoding the retinoic acid receptors a, fi, y and CRABP in the developing limbs of the mouse
.
Nature
342
,
702
705
.
Dollé
,
P.
,
Ruberte
,
E.
,
Leroy
,
P.
,
Morriss-Kay
,
G.
and
Chambón
,
P.
(
1990
).
Retinoic acid receptors and cellular retinoid binding proteins. I. A systematic study of their differential pattern of transcription during mouse organogenesis
.
Development
110
,
11331151
.
Durston
,
A. J.
,
Timmermans
,
J. P. M.
,
Hage
,
W. J.
,
Hendriks
,
H. F. J.
,
de Vries
,
N. J.
,
Heideveld
,
M.
and
Nieuwkoop
,
P. D.
(
1989
).
Retinoic acid causes an anteroposterior transformation in the developing central nervous system
.
Nature
340
,
140
144
.
Futterman
,
S.
(
1962
).
Enzymatic oxidation of vitamin A aldehyde to vitamin A acid
.
J. Biol. Chem
.
237
,
677
680
.
Geelen
,
J. A. G.
(
1979
).
Hypervitaminosis A induced teratogenesis
.
CRC Crit. Rev. Toxicol
.
6
,
351
375
.
Giguère
,
V.
,
Ong
,
E. S.
,
Segui
,
P.
and
Evans
,
R. M.
(
1987
).
Identification of a receptor for the morphogen retinoic acid
.
Nature
330
,
624
629
.
Giroud
,
A.
and
Martinet
,
M.
(
1961
).
Au sujet d’embryons cyclocéphales
.
C. R. Soc. Biol
.
155
,
2074
.
Giroud
,
A.
,
Martinet
,
M.
and
Roux
,
C.
(
1962
).
Migrations anormales des fibres optiques et considérations générales
.
Arch. Anat. Histol. Embryol
.
45
,
177
190
.
Gormley
,
P. E.
,
Rossitch
,
E.
,
D’Anna
,
M. E.
and
Cysyk
,
R.
(
1983
).
An extremely potent anilinoacridine inhibitor of aldehyde oxidase
.
Biochem. Biophys. Res. Comm
.
116
,
759
764
.
Greenfield
,
N. J.
and
Pietruszko
,
R.
(
1977
).
Two aldehyde dehydrogenases from human liver
.
Biochim. Biophys. Acta
483
,
3545
.
Hale
,
F.
(
1937
).
The relation of maternal vitamin A deficiency to microphthalmia in pigs
.
Texas State J. Med
.
33
,
228
232
.
Hlrschel-Scholz
,
S.
,
Siegenthaler
,
G.
and
Saurat
,
J.-H.
(
1989
).
Ligand-specific and non-specific in vivo modulation of human epidermal cellular retinoic acid binding protein (CRABP)
.
Eur. J. Clin. Invest
.
19
,
220
227
.
Izpisua-Belmonte
,
J. C.
,
Tickle
,
C.
,
Dollé
,
P.
,
Wolpert
,
L.
and
Duboule
,
D.
(
1991
).
Expression of the homeobox Hox-4 genes and the specification of position in chick limb development
.
Nature
350
,
585
589
.
Krust
,
A.
,
Kastner
,
P.
,
Petkovich
,
M.
,
Zelent
,
A.
and
Chambón
,
P.
(
1989
).
A third human retinoic acid receptor, hRAR-γ
Proc. Natl. Acad. Sci. USA
86
,
5310
5314
.
Kurys
,
G.
,
Ambroziak
,
W.
and
Pietruszko
,
R.
(
1989
).
Human aldehyde dehydrogenase. Purification and characterization of a third isozyme with low Km for y-aminobutyraldehyde
.
J. Biol. Chem
.
264
,
4715
4721
.
Lakshmanan
,
M. R.
,
Vaidyanathan
,
C. S.
and
Cama
,
H. R.
(
1964
).
Oxidation of vitamin Al aldehyde and vitamin A2 aldehyde to the corresponding acids by aldehyde oxidase from different species
.
Biochem. J
.
90
,
569
573
.
Lammer
,
E. J.
,
Chen
,
D. T.
,
Hoar
,
R. M.
,
Agnish
,
N. D.
,
Benke
,
P. J.
,
Braun
,
J. T.
,
Curry
,
C. J.
,
Fernhoff
,
P. M.
,
Grix
,
A. W.
,
Lott
,
I. T.
,
Richard
,
J. M.
and
Sun
,
S. C.
(
1985
).
Retinoic acid embryopathy
.
New England J. Med
.
313
,
837
841
.
Lee
,
M.-O.
,
Manthey
,
C. L.
and
Sladek
,
N. E.
(
1991
).
Identification of mouse liver aldehyde dehydrogenases that catalyze the oxidation of retinaldehyde to retinoic acid
.
Biochem. Pharmacol
.
42
,
12791285
.
Leo
,
M. A.
,
Kim
,
C.-I.
and
Lieber
,
C. S.
(
1987
).
NAD-dependent retinol dehydrogenase in liver microsomes
.
Arch. Biochem. Biophys
.
259
,
241
249
.
Lindahl
,
R.
and
Evces
,
S.
(
1984a
).
Rat liver aldehyde dehydrogenase. Isolation and characterization of four high Km normal liver isozymes
.
J. Biol. Chem
.
259
,
11986
11990
.
Lindahl
,
R.
and
Evces
,
S.
(
1984b
).
Rat liver aldehyde dehydrogenase. II. Isolation and characterization of four inducible isozymes. J. Biol. Chem
.
259
,
11991
11996
.
Lindahl
,
R.
and
Hempel
,
J.
(
1990
).
Aldehyde dehydrogenases: what can be learned from a baker’s dozen sequences?
In Enzymology and Molecular Biology of Carbonyl Metabolism III
. (Edited by
H.
Weiner
,
B.
Wermuth
and
D. W.
Crabb
) pp
1
8
.
New York
:
Plenum Press
.
Maden
,
M.
,
Ong
,
D. E.
,
Summerbell
,
D.
and
Chytil
,
F.
(
1989
).
The role of retinoid-binding proteins in the generation of pattern in the developing limb, the regenerating limb and the nervous system
.
Development
107
(
suppl
.),
109
119
.
Manthey
,
C. L.
,
Landkamer
,
G. J.
and
Sladek
,
N. E.
(
1990
).
Identification of the mouse aldehyde dehydrogenases important in aldophosphamide detoxification
.
Cancer Res
.
50
,
4991
5002
.
McCaffery
,
P.
,
Neve
,
R. L.
and
Drfiger
,
U. C.
(
1990
).
A dorso-ventral asymmetry in the embryonic retina defined by protein conformation
.
Proc. Natl. Acad. Sci. USA
87
,
8570
8574
.
McCaffery
,
P.
,
Tempst
,
P.
,
Lara
,
G.
and
Drâger
,
U. C.
(
1991
).
Aldehyde dehydrogenase is a positional marker in the retina
.
Development
112
,
693
702
.
Mendelsohn
,
C.
,
Ruberte
,
E.
,
LeMeur
,
M.
,
Morriss-Kay
,
G.
and
Chambón
,
P.
(
1991
).
Developmental analysis of the retinoic acidinducible RAR-β2 promotor in transgenic animals
.
Development
113
,
723
734
.
Napoli
,
J. L.
(
1986
).
Retinol metabolism in LLC-PK1 cells. Characterization of retinoic acid synthesis by an established mammalian cell line
.
J. Biol. Chem
.
261
,
13592
13597
.
Napoli
,
J. L.
and
Race
,
K. R.
(
1988
).
Biogenesis of retinoic acid from 0-carotene. Differences between the metabolism of β-carotene and retinal
.
J. Biol. Chem
.
263
,
17372
17377
.
Noji
,
S.
,
Nohno
,
T.
,
Koyama
,
E.
,
Muto
,
K.
,
Ohyama
,
K.
,
Aoki
,
Y.
,
Tamura
,
K.
,
Ohsugi
,
K.
,
Ide
,
H.
,
Taniguchi
,
S.
and
Saito
,
T.
(
1991
).
Retinoic acid induces polarizing activity but is unlikely to be a morphogen in the chick limb bud
.
Nature
350
,
83
86
.
Nomes
,
H. O.
,
Dressier
,
G. R.
,
Knapik
,
E. W.
,
Deutsch
,
U.
and
Gruss
,
P.
(
1990
).
Spatially and temporally restricted expression of Pax2 during murine neurogenesis
.
Development
109
,
797
809
.
Osumi-Yamashlta
,
N.
,
Noji
,
S.
,
Nohno
,
T.
,
Koyama
,
E.
,
Doi
,
H.
,
Eto
,
K.
and
Taniguchi
,
S.
(
1990
).
Expression of retinoic acid receptor genes in neural crest-derived cells during mouse facial development
.
Science
264
,
71
74
.
Pereira
,
F.
,
Rosenmann
,
E.
,
Nylen
,
E.
,
Kaufman
,
M.
,
Pinsky
,
L.
and
Wrogemann
,
K.
(
1991
).
The 56kDa androgen binding protein is an aldehyde dehydrogenase
.
Biochem. Biophys. Res. Comm
.
175
,
831
838
.
Perez-Castro
,
A. V.
,
Toth-Rogler
,
L. E.
,
Wei
,
L.-N.
and
Nguyen-Huu
,
M. C.
(
1989
).
Spatial and temporal pattern of expression of the cellular retinoic acid-binding protein and the cellular retinol-binding protein during mouse embryogenesis
.
Proc. Natl. Acad. Sci. USA
86
,
8813
8817
.
Petkovich
,
M.
,
Brand
,
N.
,
Krust
,
A.
and
Chambón
,
P.
(
1987
).
A human retinoic acid receptor which belongs to the family of nuclear receptors
.
Nature
330
,
444
450
.
Rabacchi
,
S. A.
,
Neve
,
R. L.
and
Drfiger
,
U. C.
(
1990
).
A positional marker for the dorsal retina is homologous to the 68kD-laminin receptor
.
Development
109
,
521
531
.
Rossant
,
J.
,
Zlrngibl
,
R.
,
Cado
,
D.
,
Shago
,
M.
and
Giguère
,
V.
(
1991
).
Expression of a retinoic acid response element-hsplacZ transgene defines specific domains of transcriptional activity during mouse embryogenesis
.
Genes Dev
.
5
,
1333
1344
.
Ruberte
,
E.
,
Dollé
,
P.
,
Chambón
,
P.
and
Morriss-Kay
,
G.
(
1991
).
Retinoic acid receptors and cellular retinoic binding proteins. II. Their differential pattern of transcription during early morphogenesis in mouse embryos
.
Development
111
,
45
60
.
Ruberte
,
E.
,
Dollé
,
P.
,
Krust
,
A.
,
Zelent
,
A.
,
Morriss-Kay
,
G.
and
Chambón
,
P.
(
1990
).
Specific spatial and temporal distribution of retinoic acid receptor y transcripts during mouse embryogenesis
.
Development
108
,
213
222
.
Ruiz i Altaba
,
A.
and
Jessell
,
T.
(
1991
).
Retinoic acid modifies mesodermal patterning in early Xenopus embryos
.
Genes Dev
.
5
,
175
187
.
Russo
,
J. E.
and
Hilton
,
J.
(
1988
).
Characterization of cytosolic aldehyde dehydrogenase from cyclophosphamide resistant L1210 cells
.
Cancer Res
.
48
,
2963
2968
.
Siegenthaler
,
G.
,
Saurat
,
J.-H.
and
Ponec
,
M.
(
1990
).
Retinol and retinal metabolism. Relationship to the state of differentiation of cultured kératinocytes
.
Biochem. J
.
268
,
371
378
.
Sperry
,
R. W.
(
1963
).
Chemoaffinity in the orderly growth of nerve fiber patterns and connections
.
Proc. Natl. Acad. Sci. USA
50
,
703
710
.
Strickland
,
S.
and
Mahdavi
,
V.
(
1978
).
The induction of differentiation in teratocarcinoma stem cells by retinoic acid
.
Cell
15
,
393
403
.
Sucov
,
H. M.
,
Murakami
,
K. K.
and
Evans
,
R. M.
(
1990
).
Characterization of an autoregulated response element in the mouse retinoic acid receptor type gene
.
Proc. Natl. Acad. Sci. U.S.A
.
87
,
5392
5396
.
Tabin
,
C. J.
(
1991
).
Retinoids, homeoboxes, and growth factors: toward molecular models for limb development
.
Cell
66
,
199
217
.
Theiler
,
K.
(
1972
).
The House Mouse. Development and Normal Stages from Fertilization to 4 Weeks of Age
.
Berlin, Heidelberg, New York
:
Springer-Verlag
.
Tickle
,
C.
,
Alberts
,
B.
,
Wolpert
,
L.
and
Lee
,
J.
(
1982
).
Local application of retinoic acid to the limb bud mimics the action of the polarizing region
.
Nature
296
,
564
566
.
Trisler
,
G.
,
Schneider
,
M. D.
and
Nirenberg
,
M.
(
1981
).
A topographic gradient of molecules in retina can be used to identify neuron position
.
Proc. Natl. Acad. Sci. USA
78
,
2145
2149
.
Vaessen
,
M.-J.
,
Meijers
,
J. H. C.
,
Bootsma
,
D.
and
Geurts van Kessel
,
A.
(
1990
).
The cellular retinoic-acid-binding protein is expressed in tissues associated with retinoic-acid-induced malformations
.
Development
110
,
371
378
.
Vallan
,
R. C.
and
Pietruszko
,
R.
(
1982
).
Human aldehyde dehydrogenase: mechanism of inhibition by disulfiram
.
Science
216
,
637
639
.
Wanek
,
D. M.
,
Gardiner
,
D. M.
,
Muneko
,
K.
and
Bryant
,
S. V.
(
1991
).
Conversion by retinoic acid of anterior cells into ZPA cells in the chick wing bud
.
Nature
350
,
81
83
.
Wang
,
X.-D.
,
Tang
,
G.-W.
,
Fox
,
J. G.
,
Krinsky
,
N. I.
and
Russell
,
R. M.
(
1991
).
Enzymatic conversion of β-carotene into β- apocarotenals and retinoids by human, monkey, ferret and rat tissues
.
Arch. Biochem. Biophys
.
285
,
8
16
.
Warkany
,
J.
and
Schraffenberger
,
E.
(
1946
).
Congenital malformations induced in rats by maternal vitamin A deficiency. I. Defects of the eye
.
Arch. Ophthal
.
35
,
150
169
.
Webb
,
J. L.
(
1966
).
Enzyme and Metabolic Inhibitors
.
New York and London
:
Academic Press
.
Williams
,
J. B.
and
Napoli
,
J
, L
. (
1985
).
Metabolism of retinoic acid and retinol during differentiation of F9 embryonal carcinoma cells
.
Proc. Nad. Acad. Sci. USA
82
,
4658
4662
.
Wilson
,
J. G.
,
Roth
,
C. B.
and
Warkany
,
J.
(
1953
).
An analysis of the syndrome of malformations induced by maternal vitamin A deficiency. Effects of restoration of vitamin A at various times during gestation
.
Am. J. Anat
.
92
,
189
217
.
Zelent
,
A.
,
Krust
,
A.
,
Petkovich
,
M.
,
Kastner
,
P.
and
Chambón
,
P.
(
1989
).
Cloning of murine a and f retinoic acid receptors and a novel receptor y predominantly expressed in skin
.
Nature
339
,
714717
.