INTRODUCTION

Cellular actions of vitamin A are mediated by the binding of retinoic acid to specific nuclear receptor proteins. Two families of nuclear receptors (retinoic acid receptor and retinoid X receptor) have been cloned that are active in receptor-mediated regulation of gene transcription (for review, see Ref. 1). -Carotene as a precursor to retinoic acid has been demonstrated by a number of studies, both in vitro and in vivo (for review, see Ref. 2). More recent studies (3, 4) demonstrate that 9-cis--carotene can serve as a precursor of 9-cis-retinoic acid, which is a ligand for both retinoic acid receptor and retinoid X receptor (5, 6).

Biosynthesis of retinoic acid from -carotene seems to involve both central and excentric cleavage pathways (Fig. 1). The central cleavage pathway utilizes a 15,15-dioxygenase to yield two molecules of retinal (retinaldehyde) (7, 8), which in turn can be oxidized to retinoic acid (9). The excentric cleavage mechanism would yield a series of -apocarotenals of different chain lengths (10, 11, 12). -Apocarotenals subsequently are oxidized to their corresponding -apocarotenoic acids (Fig. 1). By a process analogous to fatty acid -oxidation, the -apocarotenoic acids are converted stepwise to retinoic acid, where the -oxidation-type process is stopped by the presence of the methyl substituent on carbon 13 (Fig. 1). This hypothesis was put forth by Glover (10) and Ganguly and co-workers (13, 14). Sharma et al. (13) had isolated significant amounts of -apocarotenals as well as retinal from the intestine of chickens given -carotene. The same authors showed further that -apo-8-carotenal fed to rats was rapidly oxidized to the -apo-8, -apo-10, and -apo-12-carotenoic acids and also led to the deposition of considerable amounts of retinyl esters in the liver (14). Olson (15) indicated that -apo-8- and -apo-10-carotenals are cleaved by the mucosal dioxygenase at a markedly faster rate than -carotene. We have shown that when purified -carotene is incubated with homogenates of small intestine and other tissues, a series of homologous carbonyl cleavage products are produced including -apo-14-, -apo-12-, -apo-10-, and -apo-8-carotenal, and -apo-13-carotenone, as well as retinoic acid and retinal (-apo-15-carotenal) (16, 17). Under these experimental circumstances, the retinoic acid produced represents one-third of the total retinoids formed (2), even when citral (an inhibitor of oxidation of retinal) is added to the incubation mixture (18).

Structures of metabolites in the metabolic pathway of -carotene.

In view of these observations, we concluded that in addition to the central cleavage, an excentric cleavage mechanism is involved in the metabolism of -carotene into retinoic acid. However, the exact pathway for excentric cleavage is uncertain. Gastric tissue appears to carry out this process by using a coupled oxidation with lipoxygenase and unsaturated fatty acids (19). In the present study, we evaluate the formation of retinoic acid via a -oxidation-like process of -apocarotenoic acids by either in vitro incubation of rabbit liver mitochondrial fractions with various -apocarotenoic acids (14, 12, and 8), or by in vivo perfusion of the ferret liver with -apo-8-carotenoic acid. Since the rate of conversion of -carotene to vitamin A in rabbit intestinal mucosa in vitro is about the same in comparison with humans (2) and since this animal has been used in many -carotene metabolic experiments (20, 21), we chose rabbit liver for our in vitro incubation experiments. We chose the ferret as an animal model in the in vivo study because ferrets have many anatomic and physiological features that are similar to those of humans (22), and we have shown that the ferret is appropriate for studying the intricacies of -carotene metabolism in vivo (23). Our results suggest that one of the mechanisms for bioconversion of -apocarotenoic acid into retinoic acid is a type of -oxidation, which is reminiscent of fatty acid -oxidation.

MATERIALS AND METHODS

All-trans-retinoic acid, all-trans-retinal, all-trans-retinol, dithiothreitol, citral, dimethyl sulfoxide, Hepes, ATP, coenzyme A, dithiothreitol, NAD, FAD, L-carnitine, -tocopherol, EDTA, 3-mercaptopropionic acid (MPA),¹ and other chemicals were purchased from Sigma. 2-Tetradecylglycidic acid (TDGA, McN-3802) was a gift from the R. W. Johnson Pharmaceutical Research Institute. -Apo-14-, -apo-12-, -apo-10-, and -apo-8-carotenal, -apo-12-carotenoate, and -apo-14-, -apo-12-, and -apo-8-carotenoate esters were gifts from Hoffmann-La Roche Inc. (Nutley, NJ, and Basel, Switzerland). -Apo-14- and -apo-8-carotenoic acids were prepared by saponification of -apo-14- and -apo-8-carotenoate esters as follows: the methyl and ethyl ester of -apo-14- and -apo-8-carotenoate were saponified with 10% ethanolic KOH (pH >12) overnight at room temperature. Recovered -apocarotenoic acid was purified by the HPLC method described under ``HPLC Procedure for Analysis.'' -Apo-10-carotenoic acid was prepared by oxidation of the -apo-10-carotenal, using a procedure modified from Barua et al. (24). The -apocarotenoids were purified by HPLC to 99.8%. All HPLC solvents were obtained from J. T. Baker Chemical Co. (Philipsburg, NJ) and were filtered through a 0.45-µm membrane filter before use.

Male ferrets (Mustela putorius furo, 1.4-1.7 kg) (Marshall Farms, North Rose, NY) were fed dry ferret food (PMI, Richmond, IN). Rabbits (2-3 kg) (CAMM Research Animals, Wayne, NJ) were fed Laboratory Rabbit Diet 5326 (PMI). Eighteen surgically manipulated ferrets were used for the perfusion study.

Rabbits were euthanized by marginal ear vein injection of 0.2 g/kg, body weight, Nembutal®. Freshly isolated livers from rabbits were flushed with ice-cold Hepes buffer (buffer A; pH 7.4, 20 mM, containing 250 mM sucrose and 1 mM EDTA). The livers were homogenized on ice in a Brinkmann Polytron^TM homogenizer for 10 s at speed 10 with Hepes buffer A (w/v = 1:5) containing 1 mM dithiothreitol. The homogenate was centrifuged at 800 × g at 4 °C for 30 min. The resulting supernatant represents the crude postnuclear fraction. The supernatant fraction was collected and centrifuged by using a 70.1 Ti rotor as follows: a mitochondria-enriched fraction was obtained by centrifugation at 6000 × g for 15 min, and the mitochondrial pellet was collected and washed by rehomogenization in buffer A and recentrifuged. This mitochondrial pellet was resuspended in buffer A, adjusted to 5 mg/ml protein, and stored at 20 °C. The second supernatant was centrifuged at 100,000 × g for 60 min, with the resulting pellet labeled as the microsome-enriched fraction and the remaining supernatant fraction labeled as cytosol. Marker enzymes for subcellular fractions were analyzed (25): cytochrome c oxidase (EC 1.9.3.1) (25) for mitochondria; lactate dehydrogenase (EC 1.1.1.27) (25) for cytosol; and NADPH:cytochrome c reductase (EC 3.1.3.9) (26) for microsomes. We recovered (based on the postnuclear fraction containing 100% individual enzymatic activity) 71 ± 3% of the activity of cytochrome c oxidase, 10 ± 2% of NADPH-cytochrome c reductase and 7 ± 3% lactate dehydrogenase from the mitochondrial fraction, indicating isolation of the bulk of the mitochondria during this fractionation. Although we could decrease the contamination of the mitochondrial fraction further by using a hybrid Percoll-metrizamide gradient centrifugation (25), this procedure apparently destabilized the activity involved in -apocarotenoic acid metabolism (data not shown). Therefore, the washed mitochondria-enriched fraction was used in the in vitro incubations in all experiments. The protein concentrations of samples were determined using the BCA (bicinchoninic acid) protein assay.

Incubation of -Apocarotenoic Acids with Rabbit Liver Mitochondria-enriched Fractions

-Apocarotenoic acids (14, 12, and 8) were solubilized in dimethyl sulfoxide. The reaction mixture (final volume, 1.0 ml) contained 20 mM Hepes, pH 7.4, 30 mM KCl, 8.5 mM MgCl₂, 5 mM ATP, 0.16 mM coenzyme A, 1 mM dithiothreitol, 1 mM NAD, 0.17 mM FAD, 2.5 mM L-carnitine, 0.5 µM -tocopherol, 0.2 mM EDTA, and the carotenoid acid substrate in 10 µl of dimethyl sulfoxide. The assay was initiated by the addition of the mitochondrial fraction (100 µl, 500 µg of protein in a standard incubation). After a 60-min incubation at 37 °C, the reaction was terminated by the addition of 0.2 ml of ethanol and 20 µl of 1 N KOH to hydrolyze acyl-CoA (30 min at room temperature). The sample was extracted with 2 ml of hexane followed by the addition of 6 N HCl (40 µl) to reduce the pH to 3. A second extraction was performed with 2 ml of hexane, and the two extractions were pooled and dried under N₂. The dry residue was resuspended in 50 µl of ethanol for analysis by HPLC.

We used either our previous HPLC methods (16) or a modification thereof for assaying both -apocarotenoids and retinoids. Absorption maxima of -apo-8-, -apo-10-, -apo-12-, -apo-14-, and -apo-15-carotenoic acid (retinoic acid) were 441 nm (molar extinction coefficient  = 108,700), 424 nm ( = 81,000), 408 nm ( = 74,500), 378 nm ( = 52,200), and 340 nm ( = 44,600), respectively. To separate the -apocarotenoic acids and retinoic acid, samples were analyzed by either reverse-phase HPLC as described (16) or by a modified procedure with the pH of the HPLC mobile phase adjusted to 6.0 by adding 0.35% acetic acid, and a gradient at a flow rate of 1 ml/min was as follows: 100% solvent A (acetonitrile (CH₃CN)/tetrahydrofuran/water, 50:20:30, v/v/v, with 1% ammonium acetate) for 0.5 min, to 40% solvent B (CH₃CN/tetrahydrofuran/water, 50:44:6, v/v/v, with 1% ammonium acetate), held for 4.5 min and then linearly increased to 10% solvent A and 90% solvent B for 7 min followed by a 13-min hold and then a 2-min linear gradient to 100% solvent A, for a 3-min hold. A Waters 490E multiwavelength spectrophotometer detector was set at two different wavelengths (340 nm for retinoids and -apo-14-carotenoic acid and 400 nm for other carotenoids). In this HPLC system, retinoic acid and -apo-14, -apo-12, -apo-10, and -apo-8-carotenoic acid were separated as shown in Fig. 2C. The lower limits of detection of the assays were 1.2 pmol for retinoic acid and 2 pmol for the -apocarotenoic acids. A Waters 994 programmable photodiode array detector was used for measurement of maximum absorption. Individual carotenoids and retinoids were identified by co-elution with standards and by measurement of their absorption spectra. The compounds were quantified relative to the internal standard (retinyl acetate), by determining peak areas calibrated against known amounts of standards. The positive identification of -apocarotenoic acids and retinoic acid derived from incubations with the -apocarotenoic acids was accomplished by synthesizing and characterizing methyl carotenoates or methyl retinoate. Both the extract from the incubation and the HPLC peaks that match the retention times of standard -apocarotenoic acids and retinoic acid were suspended in 3 ml of peroxide-free diethyl ether at 4 °C. Ethereal diazomethane was then added to the suspension (27). After the reaction, the excess diazomethane was removed under nitrogen, and the residue was subjected to HPLC for direct comparison with the appropriate methyl esters of the standards.

HPLC profile of the incubation of 10 µM -apo-14-carotenoic acid with the rabbit liver mitochondrial fraction.

After an overnight fast, ketamine hydrochloride (35 mg/kg) and xylazine (3 mg/kg) were administered intramuscularly to ferrets to induce anesthesia. Ferrets were intubated with 3.0-mm inner diameter endotracheal tubes, and anesthesia was maintained with 2-3% isoflurane in 100% oxygen (North American Drager anesthesia machines, Telford, PA). Anesthetized ferrets were kept on a Homeothermic Blanket Control Unit at 38 °C monitored by an autotemperature monitor. Through a midline abdominal incision, the hepatic bile duct was cannulated using a polyethylene catheter (1.27 mm outer diameter, 0.86 mm inner diameter, PE-90). The portal vein was identified and cannulated, allowing a heparinized polyethylene tube (1.22 mm outer diameter, 0.76 mm inner diameter, PE-60) to be passed into the portal vein without obstructing portal blood flow. Normal saline was infused through the portal vein at 12 ml/h until the perfusion began. To determine the role of exogenous substrate in the processes of -apocarotenoic acid oxidation, liver perfusion was carried out in six groups (three ferrets in each group): group 1, perfusion of 2 µM -apo-8-carotenoic acid alone; group 2, co-perfusion of 2 µM -apo-8-carotenoic acid and 2 mM MPA, which is an inhibitor of long chain acyl-CoA dehydrogenase (28); group 3, perfusion of 2 µM -apo-8-carotenoic acid for the first hour followed by the addition of 2 mM MPA to the perfusate for 2 h; group 4, co-perfusion of -apo-8-carotenoic acid and 2 mM TDGA, an inhibitor of carnitine-palmitoyl-CoA transferase I (29); group 5, co-perfusion of -apo-8-carotenoic acid and 2 mM citral, which is an inhibitor of retinal oxidase and reductase (30); group 6, perfusion of 10 µM -apo-8-carotenoic acid alone. The liver perfusion solutions were prepared under red light from -apocarotenoic acid (dissolved in 2 ml of ferret plasma and normal saline, total volume of 36 ml, sonication for 15 min at 80 watts of power). The liver biopsy (about 50 mg), bile, and portal blood collected in the hour prior to experimental perfusion were saved and analyzed for base-line measurements. A syringe pump was used to infuse the solution through the portal vein at a flow rate of 12.0 ml/h in a single-pass mode. During the perfusion, bile was collected at hourly intervals, and portal vein blood was sampled via an indwelling catheter before and after the perfusion. The same volume of normal saline was simultaneously injected into the portal vein. After perfusion, the animals were killed by puncturing the abdominal aorta under deep isoflurane anesthesia. The liver was sampled and homogenized (Brinkmann Polytron^TM) with ice-cold Hepes buffer and methanol (w/v = 1 g/10 ml). The formation of metabolites was defined as the amounts appearing in bile over time using linear regression analysis (expressed as pmol/h).

The samples (serum, bile, and liver tissue) were extracted as follows: 100 µl of an ethanolic solution of 0.5 N KOH was added to either 1.0 ml of bile or serum or 50 mg of liver homogenate (in 1 ml of methanol/water (2:1, w/v) homogenized in a Brinkmann Polytron homogenizer for 10 s at speed 10), followed by the addition of the internal standard, retinyl acetate in 100 µl of ethanol. The metabolites were extracted by adding 2 ml of hexane, and the mixture was then centrifuged for 3 min at 320 × g at 4 °C. The hexane layer was removed, and the residue was acidified by adding 40 µl of 6 N HCl. A second extraction was performed with 2 ml of hexane. The two extracts were pooled, dried under N₂, and resuspended in 50 µl of ethanol for injection in the HPLC system described above. Since retinoyl -glucuronide and retinyl -glucuronide are easily hydrolyzed to retinoic acid and retinol in the presence of moderately strong acid or base (31), we did not incubate the sample with -glucuronidase before the extraction. In a preliminary study, a portion of bile was incubated with -glucuronidase (37 °C for 30 min) before the extraction, and the amounts of both retinol and retinoic acid recovered were the same as in the control incubation without -glucuronidase (data not shown). However, the involvement of glucuronidation cannot be ruled out in the perfusion study.

Results are expressed as means ± S.E., and differences among the groups were compared using Student's t test, analysis of variance, or regression analysis at p < 0.05.

RESULTS

In Vitro Metabolism of -Apocarotenoic Acid into Retinoic Acid

In an attempt to demonstrate the ``-oxidation'' reaction of -apocarotenoic acid that is similar to fatty acid -oxidation, we used various -apo-8-, -apo-12-, and -apo-14-carotenoic acids as substrates in our incubation system. Since the -oxidation-type process may be stopped by the presence of the methyl substituent on carbon 13 of retinoic acid (Fig. 1), as we expected, the terminal product of -apo-14-carotenoic acid oxidation should be retinoic acid. Rabbit liver mitochondria incubated with -apo-14-carotenoic acid produced only one peak (Fig. 2A) with a retention time (5.8 min) equivalent to the authentic standard of retinoic acid (Fig. 2C). No retinoic acid or other metabolites were detected in the control incubation using a boiled rabbit liver mitochondrial fraction (Fig. 2B). The retinoic acid (Fig. 2A, peak 1) had an absorption maximum (340 nm) that matched the authentic standard of retinoic acid (Fig. 2D). For further identification, the peak 1 fraction collected from the HPLC (Fig. 2A) was methylated by using ethereal diazomethane (5). After methylation, peak 1 appeared with a slower retention time (9.60 min), and its absorption maximum shifted to 360 nm, which matched the authentic standard of methylretinoate (Fig. 2D, inset). When 5, 10, and 20 µM -apo-14-carotenoic acids were incubated with the rabbit liver mitochondrial fraction, the rates of retinoic acid synthesis were 12 ± 1, 30 ± 7, and 34 ± 8 pmol/h/mg of protein, respectively (three separate determinations at each concentration).

When -apo-12-carotenoic acid replaced -apo-14-carotenoic acid as the substrate, we observed both the formation of -apo-14-carotenoic acid and retinoic acid (Fig. 3). The identification of -apo-14-carotenoic acid was based on its HPLC retention time and absorption maximum at 378 nm (which shifted to 394 nm after the methylation), which matched the authentic standards of -apo-14-carotenoate (or methyl -apo-14-carotenoate). Retinoic acid synthesis was linear for 60 min, while the production of -apo-14-carotenoic acid began to decline after 45 min of incubation (Fig. 3A). The incubation of -apo-12-carotenoic acid for longer times (>2 h) caused a decrease of both -apo-14-carotenoic acid and retinoic acid formation (data not shown). The formation of retinoic acid and -apo-14-carotenoic acid appeared to saturate at 20 µM -apo-12-carotenoic acid (Fig. 3B). Furthermore, -apo-8-carotenoic acid (10 µM) incubated with the mitochondrial fraction of rabbit liver gave rise to a series of shorter chain -apocarotenoic acids, as well as retinoic acid (Table I). Retinoic acid synthesis from -apo-14-, -apo-12-, and -apo-8-carotenoic acid was related linearly to the amount of protein up to 2.0 mg (Fig. 4).

Effects of incubation time (A) and substrate concentration (B) on -apo-14-carotenoic acid (filled circles) and retinoic acid (open circles) derived from -apo-12-carotenoic acid.

Table I. The formation of metabolites from various -apo-carotenoic acids at 10 µM in the mitochondrial fraction of rabbit liver The incubations were carried out as described under ``Materials and Methods.'' Data are means ± S.E. from at least three independent determinations. Substrate  -Apo-carotenoic acid products 10 12 14 15 (retinoic acid) pmol/mg protein/h  -Apo-8-carotenoic acid 35  ± 5 25  ± 2 16  ± 3 11  ± 2  -Apo-12-carotenoic acid 32  ± 4 18  ± 3  -Apo-14-carotenoic acid 30  ± 7

Effects of increasing protein concentration on retinoic acid synthesis derived from -apocarotenoic acids incubated with the mitochondrial fraction of rabbit liver.

To determine whether the microsomal fraction is necessary as a source of acyl-CoA synthetase to convert -apocarotenoic acids into their acyl-CoA derivatives, we incubated -apo-12-carotenoic acid with rabbit liver cytosol fraction, the microsomal fraction, the mitochondrial fraction, or the mitochondrial fraction plus the microsomal fraction. Less than 2 pmol of retinoic acid were detected in the incubations of -apo-12-carotenoic acid with the cytosol fraction alone or the microsomal fraction alone. There was no difference in either the -apo-14-carotenoic acid (38 ± 2 versus 35 ± 3 pmol/h) or retinoic acid formation (21 ± 2 versus 22 ± 3 pmol/h) in the incubation with 1.0 mg of mitochondrial protein or the mitochondrial fraction plus 1.0 mg of microsomal protein (means ± S.E., three separate determinations).

In view of these findings, we attempted to determine if the oxidation of the -apocarotenoic acid was at all similar to fatty acid -oxidation. The effects of oleoyl-CoA, a substrate for fatty acid -oxidation was studied at different concentrations (Fig. 5). Preincubation of mitochondria for 3 min with oleoyl-CoA caused a dose-dependent inhibition of the oxidation of -apo-12-carotenoic acid into both -apo-14-carotenoic acid and retinoic acid (Fig. 5). The concentration of oleoyl-CoA that caused 50% inhibition was close to 0.05 mM. We then studied the effect of MPA in our incubation system. The incubation of rabbit liver mitochondrial fraction with 10 µM -apo-8-carotenoic acid produced both retinoic acid and -apo-14-carotenoic acid (Fig. 6A). When we added 2 mM MPA, the formation of both retinoic acid and -apo-14-carotenoic acid was completely abolished (Fig. 6B). This inhibitory effect of MPA was dose-dependent when -apo-12-carotenoic acid was the substrate (Fig. 7). The rate of -apo-12-carotenoic acid oxidation was decreased by 80% after 3 min of preincubating mitochondria with 1.0 mM of MPA. The concentration of this inhibitor that caused 50% inhibition was close to 0.3 mM. The formation of retinoic acid from -apo-8-carotenoic acid was also completely abolished by adding 0.5 mM TDGA (Table II). Similar to our previous studies on the incubation of -carotene (18), the formation of retinoic acid from -apo-8-carotenoic acid was not inhibited by 2 mM citral (Table II). Furthermore, the addition of citral into the incubation mixture significantly increased the formation of retinoic acid from -apo-8-carotenoic acid. The oxidation of -apo-8-carotenoic acid was significantly inhibited by deletion of ATP (55%) and CoA (73%). The formation of retinoic acid from -apo-8-carotenoic acid was not affected by deleting NAD, FAD, or carnitine from the incubation buffer (Table II).

Effect of oleoyl-CoA on both the production of retinoic acid (open circles) and -apo-14-carotenoic acid (filled circles) from the incubation with 10 µM -apo-12-carotenoic acid.

The formation of retinoic acid and -apo-14-carotenoic acid from 10 µM -apo-8-carotenoic acid in the mitochondrial fraction of rabbit liver with (B) or without 2 mM MPA (A).

Effect of MPA on the production of retinoic acid (open circles) and -apo-14-carotenoic acid (filled circles) by the incubation of -apo-12-carotenoic acid.

Table II. Effects of deleting co-factors or adding inhibitors on the formation of retinoic acid from 10 µM -apo-8-carotenoic acid in the mitochondrial fraction of rabbit liver Incubation mixtures contained 20 mM HEPES, pH 7.4, 30 mM KCl, 8.5 mM MgCl2, 5 mM ATP, 0.16 mM coenzyme A, 1 mM dithiothreitol, 1 mM NAD, 0.17 mM FAD, 2.5 mM L-carnitine, 0.5 µM -tocopherol, 0.2 mM EDTA, and 10 µM -apo-8-carotenoic acid. Data are means ± S.E. from at least three independent determinations. Retinoic Acid pmol/mg protein/h Complete system 11  ± 2a Deletions  0.16 mM co-enzyme A 3  ± 1b  5 mM ATP 5  ± 2b  2.5 mM carnitine 8  ± 3a  1 mM NAD 12  ± 4a  0.17 mM FAD 10  ± 2a  All cofactors      NDd Additions  2 mM citral 16  ± 2c  0.5 mM TDGA      ND a-c Data not sharing a common superscript letter are significantly different at p < 0.05. d  ND, not detected.

In Vivo Conversion of -Apo-8-carotenoic Acid into Retinoic Acid in the Ferret

All animals were perfused continuously with normal saline for 60 min (as base line), followed by perfusion of -apo-8-carotenoic acid and various inhibitors (MPA, TDGA, or citral) for 180 min. There are no significant differences in body weight among the six groups (groups 1-6: 1.63 ± 0.09, 1.50 ± 0.05, 1.67 ± 0.12, 1.53 ± 0.03, 1.67 ± 0.07, and 1.62 ± 0.10 kg, respectively). There was no difference in the rate of bile flow among the six groups. The rate of bile production by the perfused ferret liver remained constant (0.91 ± 0.1 ml/h, range between 0.80 and 1.2 ml/h) during the liver perfusion. The observed rates in the successive 60-min intervals were 23.4, 25.4, 24.6, and 26.6% of the total volume of bile produced. The addition of MPA, TDGA, or citral did not affect bile flow.

The formation of retinoic acid in the bile after perfusion of either 2 or 10 µM -apo-8-carotenoic acid showed a constant increase through the entire perfusion period (Fig. 8), although there was no dose dependence in the appearance of retinoic acid using either 2 µM or 10 µM -apo-8-carotenoic acid. Similar to our in vitro incubation data (Fig. 6), almost complete inhibition of retinoic acid formation occurred when 2 mM MPA was added at the beginning of the -apo-8-carotenoic acid perfusion. When the MPA was added after 60 min, an immediate inhibition of retinoic acid formation was observed. The addition of 2 mM citral to the -apo-8-carotenoic acid perfusion resulted in an increase in retinoic acid formation. Furthermore, the addition of 2 mM TDGA in the -apo-8-carotenoic acid perfusate caused a 75% decrease in retinoic acid secretion in the bile.

Effect of citral, TDGA, and MPA on the production of retinoic acid from -apo-8-carotenoic acid in the perfused ferret liver, as measured by collecting retinoic acid in bile.

The perfusion of 10 µM -apo-8-carotenoic acid resulted in the accumulation of retinoic acid in the ferret liver (15 ± 1.1 pmol/100 mg of liver weight, Fig. 9). However, retinoic acid was not detected in the ferret liver before or after perfusion of 2 µM -apo-8-carotenoic acid, which may be because of the lower limits of detection of the assays. We did not detect any intermediate compounds in vivo, i.e. -apo-12 or 14-carotenoic acid, either in liver or in bile.

HPLC profile of the ferret liver (100 mg) before (A) and after (B) perfusion of 10 µM -apo-8-carotenoic acid into the portal vein.

In addition to measuring the secretion of retinoic acid in bile, we also determined the bile retinol concentration. Fig. 10 represents the change in the retinol secretion during the 3-h perfusion period, compared with the retinol secretion in the hour preceding perfusion with -apo-8-carotenoic acid. There were no significant differences between the group perfused with -apo-8-carotenoic acid alone and those perfused with the addition of citral, TDGA, or MPA. However, there was a significant difference between the group perfused with -apo-8-carotenoic acid and citral and the group perfused with -apo-8-carotenoic acid and MPA (p < 0.05); i.e. citral decreased the appearance of retinol in bile, whereas MPA enhanced retinol appearance (Fig. 10). There were no significant changes in retinol and retinyl ester concentrations in liver tissue before and after the liver perfusion (data not shown).

Effect of citral, TDGA, and MPA perfused along with -apo-8-carotenoic acid through the portal vein on the bile secretion of retinol.

DISCUSSION

Earlier studies have shown that -apocarotenals, formed from the cleavage of -carotene, are rapidly oxidized to the corresponding -apocarotenoic acids in both animals (13, 14) and humans (32). In this study, we have investigated whether the mitochondrial fraction of rabbit liver could carry out the further metabolism of -apocarotenoic acids, with particular emphasis on the formation of products indicative of oxidation at the -position of each carotenoic acid, e.g. the conversion of -apo-8-carotenoic acid to the 10-acid, the 10-acid to the 12-acid, the 12-acid to the 14-acid, and finally, the 14-acid to 15-carotenoic acid (retinoic acid) (Fig. 1). Using 10 µM -apo-8-, -apo-12-, and -apo-14-carotenoic acids as substrates (Table I), comparable rates of -apo-10-carotenoic acid, -apo-14-carotenoic acid, and -apo-15-carotenoic acid (retinoic acid) were achieved, respectively, using the rabbit liver mitochondrial fraction as the enzyme source. These results support the hypothesis that -apocarotenoic acids are oxidized in the -position, removing the successive two carbon units from the linear polyene chain. Since the rate of formation of each product shortened by two carbons is approximately the same when the individual -apocarotenoic acids were used as substrates, this would suggest that the methyl groups on carbons 9 and 13 (which are in -position; see Fig. 1) do not interfere with the -oxidation-like process of -apocarotenoic acid metabolism. However, the -oxidation-type process would be blocked by a methyl group in the -position, which occurs with the methyl group on carbon 13, thus preventing any further oxidation beyond carbon 15, i.e. forming retinoic acid as the terminal oxidation product of this -oxidation-like process of -apocarotenoic acids. This hypothesis is supported by our finding that retinoic acid formation was increased in the order of -apo-14-carotenoic acid > -apo-12-carotenoic acid > -apo-8-carotenoic acid when these carotenoic acids were incubated at 10 µM with the mitochondrial fraction (Fig. 4). Additional evidence is the finding that the stepwise oxidative products of -apo-8-carotenoic acid in mitochondria occurred in the order of -apo-10-carotenoic acid > -apo-12-carotenoic acid > -apo-14-carotenoic acid > -apo-15-carotenoic acid (retinoic acid) (Table I, Fig. 3).

The production of retinoic acid reported here is relatively small in comparison with other studies using the cytosol fractions of various rat tissues (34), but it may be difficult to reconcile in vitro rates from different systems. In our previous in vitro study (16), we could not detect any significant decrease in retinoic acid formation when citral was added to the incubation mixture of human intestinal homogenates, suggesting that the excentric cleavage pathway contributes to the production of retinoic acid from -carotene under our experimental conditions. What is probably more important is our earlier observation (33) that there was a 47% decrease in retinoic acid synthesis in the intestinal mucosa of ferrets after the in vivo perfusion of -carotene with citral (19 ± 3 pmol/g versus 36 pmol/g without citral), suggesting that both central and excentric cleavage pathways were operating to produce retinoic acid at equal amounts. Also, in the present study, there was no reduction of retinoic acid formation when citral was added either to the liver mitochondrial incubation mixture in vitro (Table II) or the perfusate in vivo (Fig. 8). In fact, the addition of citral to the in vivo perfusion system apparently increased the rate of retinoic acid appearance in bile during the first hour (Fig. 8), which may be due to the inhibition of both retinal oxidase and retinal reductase. These data offer further evidence for the biosynthesis of retinoic acid from -carotene via an excentric cleavage pathway. Napoli and Race (34) have demonstrated the ability of -carotene to serve as a precursor to retinoic acid in vitro and stated that determining whether a pathway exists for retinoic acid synthesis that does not require retinol as a substrate is a prerequisite to understanding the metabolic relationship between carotenoids and retinoids. Retinal was not detected in their studies, which suggests that -carotene was converted to retinoic acid though a pathway that did not go through retinal.

MPA is a potent inhibitor of long chain acyl-CoA dehydrogenase in coupled rat heart mitochondria (28). This interesting property prompted us to examine the effects of MPA on -apocarotenoic acid metabolism both in in vitro incubations and in perfused liver from ferrets. The results clearly show that MPA inhibited the oxidation of -apocarotenoic acids in the mitochondrial fraction (Fig. 7). In our in vitro study, 1.0 mM MPA caused 80% inhibition of -apo-12-carotenoic acid oxidation, whereas only 0.1-0.2 mM MPA is necessary to bring about the same degree of inhibition in fatty acid oxidation (28). However, 17 mM MPA is necessary for the complete inhibition of fatty acid oxidation in vivo (35), whereas we have demonstrated the immediate inhibition of retinoic acid formation after the addition of 2 mM MPA to the perfusate (Fig. 8), strongly suggesting that -apocarotenoic acid metabolism goes through a step analogous to the dehydrogenase step in the -oxidation of fatty acids.

Long chain fatty acids must first enter mitochondria by a carnitine-dependent process to be available for -oxidation, and the carnitine-palmitoyl transferase system may be rate-limiting for long chain fatty acid oxidation (36). TDGA inhibits oxidation of long chain fatty acids in the liver, by irreversibly inhibiting carnitine-palmitoyl transferase I (29). We used TDGA to test if oxidation of -apocarotenoic acid was comparable with fatty acid -oxidation. The in vivo addition of TDGA in the perfusate caused a 75% decrease in retinoic acid secretion in bile (Fig. 8). When we added TDGA into the incubation mixture of -apocarotenoic acid and mitochondria in vitro, the formation of retinoic acid was also inhibited (Table II), which indicates that a similar system might exist with respect to carotenoic acid transport into mitochondria. This hypothesis was also supported by our observation that the biosynthesis of retinoic acid declined markedly when oleoyl-CoA was added into the incubation mixture (Fig. 5). In our study, the deletion of carnitine, NAD, and FAD from our system did not result in a significant reduction in retinoic acid formation (Table II). A possible explanation for these results is that sufficient endogenous compounds are associated with the mitochondrial fraction to facilitate both carotenoic acid entry and oxidation. The addition of a microsomal fraction did not stimulate mitochondrial oxidation of -apocarotenoic acids, suggesting that whatever enzymes or cofactors are necessary to activate and facilitate carotenoic acid entry into mitochondria are contained in the mitochondrial fraction alone. Another possibility is that the mitochondrial fraction used in our incubation may contain other subcellular organelles. Although we do not have direct evidence that -apocarotenoic acid converts to -apocarotenoyl-CoA before its oxidation, the deletion of CoA resulted in a 70% inhibition, indicating that CoA plays an essential role in the oxidation of -apocarotenoic acids in mitochondria.

To demonstrate the conversion of -apocarotenoic acid into retinoic acid in the living organism, we carried out a kinetic study after infusing -apo-8-carotenoic acid through the portal vein of ferrets and monitoring retinoic acid secretion in bile and accumulation of retinoic acid in the liver. Since the increase of retinoic acid concentration in the ferret intestine and portal blood after the intestinal perfusion of -carotene has been demonstrated using chemical derivatization and gas chromatography-mass spectrometry techniques (37) and one of the major pathways of retinoic acid metabolism is though glucuronidation and secretion in the bile (38), we reasoned that retinoic acid secreted in the bile would reflect the metabolic fate of -apocarotenoic acid in the liver. Bile production is a sensitive indicator of liver function during the perfusion period, and since there was no difference in bile production among the six groups, we were able to compare directly the result from the six groups. We observed the formation of both retinoic acid and retinol from the liver perfusion of -apocarotenoic acids. It can be seen that MPA, but not citral, blocks the formation of retinoic acid (Fig. 8), whereas citral, but not MPA, decreases the appearance of retinol in the bile (Fig. 10). Therefore, these results reflect that both central and excentric cleavage pathways of -apocarotenoic acids are active in the liver (Fig. 1).

We did not detect any retinoic acid in the ferret liver before or after the perfusion of 2 µM -apocarotenoic acid, using a small biopsy sample (50 mg). Larger biopsy samples, taken before the perfusion, could damage liver function and confound subsequent results. However, after the perfusion of 10 µM -apo-8-carotenoic acid for 180 min, we detected 150 pmol retinoic acid/g liver sample (Fig. 9), equivalent to 4500 pmol/whole liver (based on the liver weight of 30 g/1.5-kg ferret (22)). During this same 3-h perfusion period, we collected 90 pmol (Fig. 8) of retinoic acid in the bile, indicating that only a very small fraction of the newly formed retinoic acid is secreted in the bile and that the -oxidation-like pathway of the -apocarotenoic acids may contribute to the steady-state concentration of retinoic acid in the body.