Lecithin:Retinol Acyltransferase Is Responsible for Amidation of Retinylamine, a Potent Inhibitor of the Retinoid Cycle*

Lecithin:retinol acyltransferase (LRAT) catalyzes the transfer of an acyl group from the sn-1 position of phosphatidylcholine to all-trans-retinol (vitamin A) and plays an essential role in the regeneration of visual chromophore as well as in the metabolism of vitamin A. Here we demonstrate that retinylamine (Ret-NH2), a potent and selective inhibitor of 11-cis-retinal biosynthesis (Golczak, M., Kuksa, V., Maeda, T., Moise, A. R., and Palczewski, K. (2005) Proc. Natl. Acad. Sci. U. S. A. 102, 8162-8167), is a substrate for LRAT. LRAT catalyzes the transfer of the acyl group onto Ret-NH2 leading to the formation of N-retinylpalmitamide, N-retinylstearamide, and N-retinylmyristamide with a ratio of 15:6:2, respectively. The presence of N-retinylamides was detected in vivo in mice supplemented with Ret-NH2. N-Retinylamides are thus the main metabolites of Ret-NH2 in the liver and the eye and can be mobilized by hydrolysis/deamidation back to Ret-NH2. Using two-photon microscopy and the intrinsic fluorescence of N-retinylamides, we showed that newly formed amides colocalize with the retinyl ester storage particles (retinosomes) in the retinal pigment epithelium. These observations provide new information concerning the substrate specificity of LRAT and explain the prolonged effect of Ret-NH2 on the rate of 11-cis-retinal recovery in vivo.

In vertebrates, the retinoid cycle is essential for regeneration of the chromophore 11-cis-retinal, which is an integral part of rhodopsin and cone visual pigments (1). All-trans-retinol is generated by the photoisomerization of 11-cis-retinal bound to opsins in photoreceptor cells or supplemented from circulation. It is trapped in the retinal pigment epithelium (RPE) 3 in the form of insoluble fatty acid esters in subcellular structures known as retinosomes (2,3). The enzyme responsible for the esterification of all-trans-retinol in the small intestine, liver, and eye, lecithin:retinol acyltransferase (LRAT) (4,5), was cloned (6), and its function has been proven in vivo (2,7). Later in the retinoid cycle the 11-cis configuration of the retinal is restored by enzymatic isomerization (8).
The key step in the transformation of all-trans-retinal to 11-cis-retinal is the isomerization reaction. Recently a candidate protein approach and expression cloning demonstrated that RPE65 exhibits Fe 2ϩdependent isomerization activity (9 -11). We have suggested that the regeneration of the chromophore might occur through a retinyl carbocation intermediate (12) and demonstrated that isomerization is inhibited by positively charged retinoids (13). This mechanism would be consistent with Fe 2ϩ -catalyzed alkyl cleavage of the retinyl esters. Retinylamine (Ret-NH 2 ) potently and selectively inhibits the isomerization step of the retinoid cycle in vitro and in vivo, whereas modifications of the amino group lead to loss of inhibitory potency. Surprisingly Ret-NH 2 has a long lasting effect, and when mice were treated with a single dose of the inhibitor, its inhibitory effect on the cycle was observed for several days (13).
Inhibition of the retinoid cycle may have implications in averting light damage to photoreceptors in some instances or preventing the accumulation of toxic condensation side products of bleached chromophore, all-trans-retinal, in the RPE. In Stargardt disease, a disease associated with mutations in the photoreceptor-specific ATP-binding cassette transporter (ABCR) (14) or elongation of the very long chain fatty acidlike 4 protein (ELOVL 4) (15,16), the accumulation of all-trans-retinal is thought to be responsible for the formation of a component of a lipofuscin pigment called A2E. This fluorophore is accumulated during the course of the disease and, to a lesser extent, with age in normal individuals. It exerts a toxic effect on retinal cells, causing retinal degeneration and consequent loss of vision (17,18). In contrast with other less potent inhibitors such as 13-cis-retinoic acid or its metabolically active intermediates (19), Ret-NH 2 does not activate the transcription of genes (13), making it a safer candidate for therapeutic application.
However, a total inhibition of 11-cis-retinoid production would resemble Leber congenital amaurosis (LCA), an autosomal recessive rod-cone dystrophy that presents itself at birth or the first few months of life. Rod and cone photoreceptor functions of the LCA patients are either absent or severely compromised at birth as evidenced by extinguished or barely detectable photopic and scotopic electroretinograms (20). LCA is highly heterogeneous; at least nine subtypes of LCA have been identified (see www.sph.uth.tmc.edu/RetNet/disease.htm) and linked to many unrelated genes. For enzymes of the retinoid cycle, numerous mutations in the RPE65 gene have been linked to LCA (21)(22)(23)(24). Null mutations in the human LRAT and RDH12 genes lead to recessive early onset retinal dystrophy, a phenotype very similar to LCA (25)(26)(27). Thus, a reduced but not blocked regeneration of rhodopsin could be beneficial for some forms of retinal dystrophies, particularly for a low dose and long lasting inhibitor.
Here we provide evidence that Ret-NH 2 is converted to pharmacologically inactive retinylamides in vitro and in vivo in the liver and RPE. We demonstrate that the enzyme responsible for the amidation is LRAT and that the amidation products are stored in retinosomes in the RPE.
N-retinylpalmitamide were prepared by reacting Ret-NH 2 with an excess of acetic anhydride and palmitoyl chloride, respectively, in anhydrous dichloromethane in the presence of N,N-dimethylaminopyridine at 0°C for 1 h. After the reaction was completed (as judged by HPLC), water was added, and the product was extracted with hexane. The hexane layer was washed with saturated NaCl solution, dried with anhydrous magnesium sulfate, filtered, and evaporated in a SpeedVac. Radioactive Ret-NH 2 and N-retinylamides were synthesized starting from 11,12-[ 3 H]all-trans-retinol, which was first oxidized with MnO 2 (CH 2 Cl 2 , 20°C, 4 h) to 11,12-[ 3 H]all-trans-retinal. Ret-NH 2 and retinylamides were then prepared using methods described previously (13). N-Retinylheptanamide was prepared by dicyclohexylcarbodiimide-promoted coupling reaction. First dicyclohexylcarbodiimide was reacted with heptanoic acid in dichloromethane. Ret-NH 2 in dichloromethane was added. The reaction mixture was incubated at room temperature for 3 h, extracted, and washed as described above. Mass spectrometry (MS) of synthesized retinoids was performed using a Kratos Analytical Instruments HV-3 direct probe mass spectrometer and electron-impact ionization.
Inducible Expression of LRAT Protein in HEK Cells-Mouse LRAT cDNA was cloned as described elsewhere (7). For expression, the LRAT coding region was amplified using the primers 5Ј-GCCACCATGAA-GAACCCAATGCTGGAAGCT-3Ј and 5Ј-ACATACACGTTGAC-CTGTGGACTG-3Ј. The PCR product was ligated into the pCR-Blunt II-TOPO vector (Invitrogen) and then subcloned into the EcoRI site of pcDNA4/TO. N-Acetylglucosaminyltransferase I-negative HEK293S cells (31), obtained from Dr. G. Khorana (Massachusetts Institute of Technology, Boston, MA), were transfected with the TetR expression plasmid pcDNA6-TR(blaR), and blasticidin-resistant colonies were selected and cloned. Cells were cultured in Dulbecco's modified Eagle's FIGURE 1. Incubation of Ret-NH 2 with RPE microsomes leads to formation of a less polar product. A, Ret-NH 2 (20 M) was incubated with bovine RPE microsomes in 37°C for 2 h, extracted, and examined by normal phase HPLC (0.5% ammonia/methanol in ethyl acetate). Chromatogram reveals the presence of newly formed products (1, 2, and 3) characterized by shorter elution time (top panel). By changing elution conditions (10% ethyl acetate/hexane), three products of the reaction can be separated (bottom panel). All of them exhibit absorbance spectra indistinguishable from Ret-NH 2 . B, the rate of formation of the new product reveals that the reaction is significantly slower than the esterification of all-transretinol by LRAT. mAU, milliabsorbance units. medium, 10% fetal calf serum, and Zeocin and blasticidin antibiotics, and maintained at 37°C, 5% CO 2 , and 100% humidity. TetR-expressing HEK cells were transfected with the pcDNA4/TO-LRAT construct and selected with Zeocin. Stable clones were verified for expression of LRAT protein using the anti-LRAT monoclonal antibody (7).
LRAT Activity Assay-The reaction was carried out in 10 mM bis-Tris propane buffer, pH 7.5, 1% bovine serum albumin. All-trans-retinol or Ret-NH 2 was delivered in 0.8 l of N,N-dimethylformamide to the final concentration of 20 M. The reaction was initiated by adding 20 l of bovine RPE microsomes or 50 l of LRAT-expressing HEK cell lysate (ϳ150 mg of protein). The total volume of the reaction mixture was fixed at 200 l. The reactions were incubated at 37°C for the required times and then stopped by adding 300 l of methanol followed by the same volume of hexane. Retinoids were extracted and analyzed on a Hewlett Packard 1100 series HPLC system equipped with a diode array detector. A normal phase column (Beckman Ultrasphere-Si, 5 m, 4.5 ϫ 250 mm) and a step gradient of ethyl acetate in hexane at a flow rate of 2 ml/min were used to elute N-retinylamides (10% ethyl acetate for 23 min and then 40% ethyl acetate up to 40 min). To detect Ret-NH 2 , retinoid separation was performed in 99.5% ethyl acetate with the addition of 0.5% of 7 N ammonia in methanol.
Mouse Retinoid Extraction and Analysis-Two mouse eyes or 0.5 g of mouse liver were homogenized in a glass-to-glass homogenizer using 3 ml of 50% methanol in 20 mM bis-Tris buffer, pH 7.4. Retinoids were extracted with 4 ml of hexane. The organic phase was collected, dried down in a SpeedVac, and redissolved in 400 l of hexane. In the case of the liver extract, 10 and 100 l of retinoid solution were injected on an HPLC column for the detection of N-retinylamides and Ret-NH 2 , respectively. For samples from mouse eyes, 100 l were analyzed for N-retinylamides and Ret-NH 2 . Separation conditions used for retinoid analysis were the same as described above. To determine the radioactivity distribution among retinoids found in the liver of animals gavaged with 11,12-[ 3 H]alltrans-Ret-NH 2 or 11,12-[ 3 H]N-all-trans-retinylamide, products corresponding to the retinyl esters, retinol, retinylamides, and Ret-NH 2 were collected during an HPLC run. Fractions containing retinoids were dried down in a SpeedVac and redissolved in 300 l of N,N-dimethylformamide. The radioactivity of each fraction was examined by scintillation counting and normalized to total radioactivity of the sample injected on the column.
Two-photon Vitamin A Imaging-Two-photon excitation microscopy was performed using a confocal/two-photon laser scanning A, the HPLC-purified biosynthetic products of the Ret-NH 2 reaction are compared with RPN for their retention time by normal phase HPLC. Elution times for the most abundant biosynthetic product (a, peak 2) and RPN (b) are identical for the examined conditions. The absorbance spectra for the two compounds are identical with a maximum absorbance at 325 nm (c). B, electron impact MS analysis of the biosynthetic product (peak 2) and RPN are identical with the parent ion at 523 m/z and two main fragments at 268 and 255 m/z (B, 2). MS of two other products labeled 1 and 2 (A, a) revealed a fragmentation pattern characteristic for retinoids (268 and 155 m/z) and masses 551 and 495, respectively (B, 1 and 3). Taking into consideration elution times, spectral properties, and masses, the products correspond to all-trans-Nretinylamides composed of C 18 and C 14 acyl groups. mAU, milliabsorbance units. DECEMBER 23, 2005 • VOLUME 280 • NUMBER 51 microscope (LSM 510 MP-NLO; Carl Zeiss MicroImaging, Inc., Thornwood, NY) with LSM510 software version 3.0. Detailed methods were described in previous publications (2,3). For localization of Ret-NH 2 in the RPE cells, eyecup preparations were exposed to 0.25 mM Ret-NH 2 caged with 100 mM (2-hydroxypropyl)-␤-cyclodextrin for 15 min and washed briefly with Ames' medium (Sigma) for 3 min. Ames' medium was equilibrated with argon to purge O 2 from the solution. WT mice were gavaged with 1 mg of Ret-NH 2 18 h prior to analysis. Retinoids were extracted from the liver and eyes and examined by HPLC. Based on the conditions and elution time, N-retinylamides (products 2, 2Ј, 3, 3Ј, and 4) were identified in the liver and eyes. Products 5 and 5Ј correspond to isomers of Ret-NH 2 (insets). The peak labeled 1 is all-trans-retinol. Comparison of the amount of Ret-NH 2 and N-retinylamides found in the tissues indicates that the majority of Ret-NH 2 is converted into amides. mAU, milliabsorbance units.

Ret-NH 2 Is Converted into N-Retinylamides upon Incubation with RPE
Microsomes-HPLC analysis of the retinoids extracted from the standard isomerization assay (12) showed no significant changes (within 5%) in the amount of Ret-NH 2 . When Ret-NH 2 was incubated with RPE microsomes in the absence of all-trans-retinol, the level of Ret-NH 2 dropped significantly with time concomitantly with the appearance of a new peak that eluted with the front of the HPLC column (Fig. 1A, top panel). Putative products of Ret-NH 2 conversion were separated in 10% ethyl acetate/hexane, revealing the presence of three different compounds (Fig. 1A, bottom panel). Based on the elution time from a normal phase HPLC column and the shape and maximum of UV absorbance spectra, these products were retinoids less polar than Ret-NH 2 (Fig. 1A). As measured by the disappearance of Ret-NH 2 , the observed reaction progress was much slower than the esterification of all-trans-retinol driven by LRAT (Fig. 1B).
The unknown compounds 1, 2, and 3 were separated and purified by collecting the appropriate fractions from a normal phase HPLC column (Fig. 1A). The purity of biosynthetic products was examined by HPLC ( Fig. 2A, a) prior to MS analysis. The biosynthetic compounds have an m/z of 551, 523, and 495 with respect to their order of elution from an HPLC column (Fig. 2B, a, b, and c). These observed masses correlate with the masses of N-retinylamides possessing C 18 , C 16 , and C 14 acyl groups within their structure. To collect more evidence, the MS pattern of synthetic N-retinylpalmitamide was compared with the most abundant biosynthetic product ( Fig. 2A, a, peak 2). The molecular ion peak and MS fragmentation patterns of both compounds are identical. Additionally comparison of elution time from the HPLC column and UV absorbance spectra revealed no differences between synthetic and biosynthetic products ( Fig. 2A, b and c). Thus, we conclude that RPE microsomes converted Ret-NH 2 into three main amides: RPN, N-retinylstearamide, and N-retinylmyristamide produced in a ratio of 15:6:2, respectively.
N-Retinylamides Can Be Detected in Tissues of Mice Gavaged with Ret-NH 2 -The potential application of Ret-NH 2 as an inhibitor of the retinoid cycle in vivo led us to investigate whether N-retinylamides can be detected in mouse tissue. Mice were gavaged with 1 mg of Ret-NH 2 18 h prior to retinoid analysis. The presence of N-retinylamides in the liver and eye extracts could be easily detected by normal phase HPLC chromatography (Fig. 3). N-Retinylamides were recognized based on comparison of their UV absorbance spectrum and elution time. Interestingly, the amount of Ret-NH 2 found in the examined tissues was smaller than the amount of N-retinylamides, 4 and 15 nmol, respectively, in the livers and 102 and 140 pmol found in the eyes (Fig. 3, insets). These results suggest that Ret-NH 2 is efficiently converted into N-retinylamides in vivo.
Ret-NH 2 Is Amidated Due to LRAT Activity-To identify the enzyme that is responsible for amidation, we tested whether LRAT can utilize Ret-NH 2 as a substrate. First we carried out qualitative analysis of retinyl esters and N-retinylamides formed in an excess of 1,2-diheptanoyl-snglycero-3-phosphocholine (DHPC). DHPC, present in millimolar concentration in the reaction mixture containing RPE microsomes and all-trans-retinol, serves as a donor of the C 7 acyl group that is efficiently transferred onto all-trans-retinol. Consequently retinyl heptanoate was produced and was identified based on its mass spectrum and elution profile from normal phase HPLC in comparison with a synthetic standard (Fig. 4A). The same experiment performed in the presence of Ret-NH 2 led to the formation of an amide containing a short acyl chain whose mass and elution time from the normal phase HPLC column perfectly matched the properties of N-retinylheptanamide (Fig. 4B). Considering transfer of the heptanoyl group from DHPC onto both all-trans-retinol and Ret-NH 2 , we conclude that LRAT could be responsible for both enzymatic activities.
In addition, the production of N-retinylamides in LRAT-expressing HEK cells was investigated. Prior to the experiments, LRAT-expressing HEK cells were examined for expression and activity of the enzyme by immunoblotting and standard assays (Fig. 5B, a and b). Cells were harvested and homogenized, and the lysate was incubated in the presence of Ret-NH 2 . The products of the reaction were analyzed by normal phase HPLC. This analysis revealed striking differences between the retinoid composition extracted from untransfected and LRAT-expressing cells. In the presence of LRAT, products corresponding to N-retinylamides were observed (Fig. 5A, top panel) that correlate with the decrease of Ret-NH 2 to below detection limits (Fig. 5A, bottom panel). In the control cells, the unreacted substrate was observed, and no amides were formed (Fig. 5A). Quantification of the retinoids found in the examined samples is shown in Fig. 5B, c. Additionally incubation of the cell lysate with Ret-NH 2 led to an elevation of all-trans-retinol that was observed directly by HPLC in untransfected cells. In the case of LRAT-expressing cells, the amount of retinyl esters increased. Thus, as an alternative to amidation, Ret-NH 2 can be also deaminated to all-trans-retinol. To further investigate the role of LRAT in Ret-NH 2 acylation in vivo, we designed assays using LratϪ/Ϫ mice. Following gavage of LratϪ/Ϫ mice with synthetic Ret-NH 2 , we observed a complete lack of N-retinylamide formation in the examined livers (Fig. 6, top) concomitant with the detection of intact Ret-NH 2 (Fig. 6, bottom). Together these observations demonstrate that LRAT utilizes Ret-NH 2 as an acceptor of acyl groups. Additionally analysis of the chromatograms obtained from in vivo studies indicated that Ret-NH 2 was transformed into vitamin A. This transformation was confirmed by the increased level of all-transretinol found in LratϪ/Ϫ mice gavaged with Ret-NH 2 as well as the elevation of retinyl esters despite the absence of LRAT (Fig. 6, top). Without LRAT, retinol esterification most likely occurs due to the action of acyl CoA:diacylglycerol acyltransferase (DGAT) enzymes (32,33) or the recently described product of the GS2 gene (34).
N-Retinylamides Can Be Hydrolyzed Back to Ret-NH 2 in Vivo-In contrast to Ret-NH 2 , N-retinylamides do not inhibit the formation of 11-cis-retinol in vitro (13). We showed that Ret-NH 2 has a long lasting affect on visual recovery (more than 48 h), which is surprising considering the efficient conversion of Ret-NH 2 into N-retinylamides observed in vivo. It may be that once synthesized, N-retinylamides do not remain intact but are slowly hydrolyzed back to Ret-NH 2 . To address this hypothesis, WT mice were gavaged with 1 mg of RAN 18 h prior to HPLC analysis of retinoids extracted from liver. Based on the retention time from a normal phase HPLC column, RAN and C 18 , C 16 , and C 14 N-retinylamides were identified (Fig. 7A). The presence of   DECEMBER 23, 2005 • VOLUME 280 • NUMBER 51 JOURNAL OF BIOLOGICAL CHEMISTRY 42269 N-retinylamides that were different from those administrated suggests that RAN is hydrolyzed to Ret-NH 2 , which is then converted primarily into RPN. Alternatively, but in our opinion less likely, is the explanation that amides can undergo a transacylation reaction.

Amidation of Retinylamine
To follow the metabolites of Ret-NH 2 and its amides, radiolabeled retinoids were used. WT mice were gavaged with 11,12-[ 3 H]Ret-NH 2 , RAN, or RPN, and after 18 h the main retinoids present in the liver were separated by HPLC, collected, and examined by scintillation counter. The distribution of radioactivity among retinoids is shown in Fig. 7B. Because Ret-NH 2 and RAN are converted mostly to RPN, it is no surprise that the radioactivity profiles for each compound look similar. Interestingly, besides RPN, most of the radioactivity was found in the fractions corresponding to the retinyl esters, reflecting the deamidation of Ret-NH 2 observed earlier. These findings led us to conclude that retinyl esters and RPN are the main metabolites of Ret-NH 2 in vivo.
The data described above encouraged us to use non-inhibitory N-retinylamides as precursors of Ret-NH 2 (a potent inhibitor of retinoid isomerase). For this purpose WT mice were gavaged with 2 mg of RAN or RPN, exposed to strong light, and left for recovery of the rhodopsin chromophore. The quantity of 11-cis-retinal in the eyes was then examined by HPLC. The experimental data revealed that animals supplemented with N-retinylamides, like with free Ret-NH 2 or its salt, exhibited a much slower recovery of 11-cis-retinal compared with untreated animals (Fig. 8).
Two-photon Imaging of Ret-NH 2 and N-Retinylamide in the RPE Cells -RPE cells, where the enzymatic process of 11-cis-retinal regeneration occurs, are targets for the biological activity of Ret-NH 2 . Using a two-photon microscopy technique (2, 3), we investigated the subcel-lular localization of Ret-NH 2 and N-retinylamides within mouse RPE cells. Albino mice gavaged with Ret-NH 2 revealed increased basal fluorescence intensity compared with control animals. This was attributed to the accumulation of retinyl esters and N-retinylamides in the eye (Fig. 9, A and D; see Fig. 3 for quantification). The distribution of fluorescence correlated with the retinoid intrinsic fluorescence pattern described previously (2) that may suggest that N-retinylamides colocalize with retinosomes. Bleaching of rhodopsin in the mouse eye led to temporary accumulation of retinyl esters (35) and, consequently, to a further increase in fluorescence within the RPE layer (Fig. 9, B and E). In the presence of Ret-NH 2 the high level of fluorescence persisted as a result of retinoid cycle inhibition (Fig. 9, C and F) in contrast with its short duration in untreated mice.
To more accurately track the accumulation and localization of N-retinylamides we switched to a less complex, more reduced system. Visualization of the retinyl esters in the isolated, unstained RPE cells displayed the characteristic pattern of retinosomes located on the peripheries of the RPE cells (Fig. 10A). To record the fluorescence signal from only exogenously delivered Ret-NH 2 , internal fluorescence emanating from endogenous retinoids was depleted by UV photobleaching. This led to the decomposition of endogenous retinoids, and consequently, no fluorescence signal could be detected using the given experimental conditions (Fig. 10B). Ret-NH 2 was then applied to the RPE cells and imaged by two-photon microscopy. A significant increase in fluorescence was observed in retinosomes as well as in the cytoplasmic/endoplasmic reticulum area. Analysis of the picture revealed that Ret-NH 2 and N-retinylamides localized to the retinosomes (Fig. 10C). By taking advantage of the pH-dependent fluorescence of Ret- NH 2 (Fig. 11) it was possible to discriminate between signals from amine and amide. Ret-NH 2 was quenched and washed out in the cytoplasmic/ endoplasmic reticulum area by Ames' medium with 0.1% Triton X-100 at pH 5.5. In this case, the retinosomes still retained fluorescence (Fig. 10D). To prove that the collected images depict N-retinylamide fluorescence, retinoids from the eyecup used for the experiment were extracted and analyzed by HPLC. Only N-retinylamides were detected in the examined samples (Fig. 10). Based on the results described above we conclude that fluorescence observed in retinosomes correlates with the presence of retinyl amides.

DISCUSSION
Ret-NH 2 is an important tool for study of the retinoid cycle and is also a potential pharmacological agent. To understand the long lasting effects of Ret-NH 2 , we analyzed the metabolism of this compound using in vitro and in vivo assays. The results revealed that Ret-NH 2 , which actively and potently inhibits the synthesis of 11-cis-retinal (13), is in equilibrium between its free and amidated forms in the liver and in the eye. LRAT is the enzyme responsible for its amidation (Fig. 12).
Transformation of Ret-NH 2 to Retinylamides and Storage in Retinosomes-Ret-NH 2 was amidated ϳ50 -100 times more slowly than all-trans-retinol was esterified to retinyl esters (Fig. 1B). C 16 was the most abundant form of the amide followed by C 18 and C 14 (Fig. 2). Several candidate enzymes could be responsible for the amidation process. Examples include fatty acid acyltransferases, which are integral membrane proteins that contain a DHHC-Cys-rich domain and are involved in protein palmitoylation (36,37), and DGAT enzymes. Recently DGAT1 was found to be involved in retinyl ester synthesis (32,33). Here we provide evidence that the enzyme responsible for the majority of Ret-NH 2 amidation is LRAT (Figs. 4 -6). Changes of the maximum fluorescence emission wavelength for retinyl palmitate, RPN, and Ret-NH 2 (480 nm) were monitored in aqueous solutions at a wide pH range. Retinoids in N,N-dimethylformamide were delivered into a measurement cell to a final concentration of 1 M for ester and amide or 5 M in the case of Ret-NH 2 . The excitation wavelength was fixed at 335 nm. Experimental data reveal that, unlike retinyl ester and amide, the fluorescence intensity of Ret-NH 2 exhibits a strong decrease upon acidification. LRAT has a modest substrate specificity in terms of recognition of the ␤-ionone moiety (38 -40). Its efficient utilization of amine as compared with an alcohol group was unexpected. However, if the mechanism of LRAT action involves first the acyl transfer from a lipid to a Cys residue at the active site of the enzyme as has been suggested (41), either alltrans-retinol or Ret-NH 2 could undergo this transacylation as a substrate.
Oxidation of Ret-NH 2 to Vitamin A-Using radioactive Ret-NH 2 , we found that this compound can also be oxidized to retinol (vitamin A) and in turn be esterified to retinyl esters (Figs. 7 and 12). Thus, the metabolism of Ret-NH 2 involves either the formation of more chemically inert amides or oxidation to vitamin A, making it an attractive compound for in vivo studies of the retinoid cycle and for potential applications to treat human retinal dystrophies.
Implication of Potential Therapeutic Use of Ret-NH 2 -The use of potent inhibitors of isomerization in vivo after light exposure would prevent or slow down the recovery of the visual pigment chromophore and could have an important protective role in cases of damagecausing excessive exposure to light. Because Ret-NH 2 inhibits the regeneration of rhodopsin, which is responsible for causing damage in response to light overexposure (42), it may be able to inhibit retinal damage in such cases. In Stargardt disease (14), associated with mutations in the ABCR transporter, the accumulation of alltrans-retinal has been proposed to be responsible for the formation of a lipofuscin pigment, A2E, which can be toxic toward retinal cells and causes retinal degeneration and consequent loss of vision (17,18). It was proposed that treating patients with an inhibitor of retinol dehydrogenases, 13-cis-retinoic acid (Accutane, Roche Applied Science) might prevent or slow down the formation of A2E and might have protective properties that could help to maintain normal vision (19,43). It is not yet clear whether Ret-NH 2 will have a protective effect in either case, but in contrast with 13-cis-retinoic acid, which can spontaneously isomerize to the all-trans isomer and in turn activate the nuclear receptors retinoid X receptor and retinoic acid receptor, Ret-NH 2 does not interact at micromolar concentrations with retinoid X receptor and retinoic acid receptor (13). Thus, Ret-NH 2 thus potentially offers a safer alternative to 13-cis-retinoic acid as a pharmacological agent in preventing or reducing damage due to light exposure and in cases of retinal degenerative disease. Ret-NH 2 is also stored in the eye and liver as retinylamides and released after hydrolysis back to free Ret-NH 2 . Importantly, Ret-NH 2 could be used in patients less frequently than 13-cis-retinoic acid.
Application of Ret-NH 2 in Studies of the Retinoid Cycle-In the RPE, retinylamides are synthesized from exogenous Ret-NH 2 , transported, and stored in retinosomes (Fig. 9). The fact that retinylamides have a long lasting inhibitory effect on the recycling of all-trans-retinal to 11-cis-retinal suggests that retinosomes, as predicted from their resemblance to lipid droplets (2,3), are active organelles involved in the retinoid cycle. Ret-NH 2 and other amides offer numerous research applications to be explored. A particularly useful application may be the pH dependence of the intrinsic fluorescence of Ret-NH 2 to discriminate the drug from other autofluorescence in the eye. Two-photon microscopy used to study the retina (2, 3) should also provide further useful applications in observing drug accumulation in the eye.