INTRODUCTION

Hydrolysis of retinyl esters occurs during both the hepatic uptake of newly absorbed dietary vitamin A and during the mobilization of retinyl ester stores from the liver. Thus, hepatic enzymes catalyzing the hydrolysis of retinyl esters are important in the body's overall vitamin A homeostasis. Earlier studies focused on a neutral, bile salt-dependent retinyl ester hydrolase that is now understood to be the enzyme carboxylester lipase (see Ref. 1 for a review). This enzyme is secreted by the pancreas into the intestinal lumen, where it can presumably hydrolyze dietary retinyl esters and other lipid esters (2-5). The enzyme is also secreted by the liver (6), and its role in hepatic retinyl ester metabolism is under investigation but is presently unclear. Carboxylester lipase has been cloned from several species, including the rat (see Ref. 7 for a review). In addition to carboxylester lipase, a number of tissues contain bile salt-independent retinyl ester hydrolase activities that have as yet not been fully purified or characterized biochemically (4, 8-10). For example, we have previously demonstrated the occurrence of both acid and neutral retinyl ester hydrolase activities associated with rat liver microsomal preparations enriched in plasma membranes and endosomes (4, 10). We recently reported studies demonstrating the co-localization of newly delivered chylomicron retinyl esters and bile salt-independent retinyl ester hydrolase enzyme activities in the same plasma membrane/endosome fractions, and the in vitro hydrolysis of chylomicron remnant retinyl esters by these fractions (11). These studies suggested a probable role for these enzymes in the initial hepatic metabolism of chylomicron retinyl esters.

In the present investigations, we have undertaken the purification and characterization of a neutral, bile salt-independent retinyl ester hydrolase from rat liver microsomes. The isolated enzyme has an apparent molecular mass of 66 kDa and catalyzes the hydrolysis of retinyl esters at higher rates than triglyceride hydrolysis. It shows no cholesteryl ester hydrolase activity. Partial amino acid sequence analysis and immunoblot analysis show that the enzyme is highly related to or identical to rat liver carboxylesterase ES-2.

EXPERIMENTAL PROCEDURES

Cholesteryl [1-¹⁴C]oleate (1.96 GBq/mmol) and [1-¹⁴C] palmitic acid (2.22 GBq/mmol) were purchased from Amersham; [carboxyl-¹⁴C]triolein (4.1 GBq/mmol) was from NEN Life Science Products; all-trans-retinol was from Sigma; octyl--D-thioglucopyranoside (OSGP)¹ was from Calbiochem; DEAE-Sepharose CL-6B, Phenyl-Sepharose CL-4B, Sephadex G-100, and Sephacryl S-200 were from Pharmacia (Uppsala, Sweden). All chemicals used for sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) were from Fisher Scientific; polyvinylidene difluoride membrane was from Bio-Rad. All other reagents were of the highest purity commercially available. Retinyl [1-¹⁴C]palmitate was synthesized as described by Prystowsky et al. (12) and was purified by chromatography on small columns of neutral alumina.

Male Sprague-Dawley rats (Ace Animal Inc., Boyertown, PA), weighing 150-250 g, were fed a commercial diet and water ad libitum. The animals were cared for in an American Association for the Accreditation of Laboratory Animal Care (AAALAC)-certified animal care facility at the Medical College of Pennsylvania-Hahnemann School of Medicine.

Rat liver microsomes were isolated as described previously (4) using the procedure developed by Amar-Costesec et al. (13).

Retinyl ester hydrolase activity was measured by the radioisotopic method as described in detail previously (4). The assay was modified by adding a final concentration of 1% OSGP and 0.1 mM bovine serum albumin per tube. The substrate used was retinyl [¹⁴C]palmitate (4 nmol and 0.05 µCi per assay, unless otherwise specified). In each assay, blank tubes (without enzyme) were measured to correct for spontaneous hydrolysis of substrate. An enzyme unit was defined as 1 pmol of retinyl [1-¹⁴C]palmitate hydrolyzed/min at 37 °C. Enzyme activity was expressed as units/mg or units/ml. Cholesteryl oleate and triolein hydrolysis activities were measured in the same assay system, using cholesteryl [1-¹⁴C]oleate and [carboxyl-¹⁴C]triolein as substrate, respectively.

To solubilize the microsomal membrane-bound NREH, a 25% OSGP (w/v) stock solution was added slowly, with constant stirring, into freshly thawed microsomal pellet suspension (~12 mg/ml) to a final concentration of 1% OSGP (w/v). Additionally, a 1 M Tris-HCl buffer (pH 7.0) was added to give a final concentration of 10 mM Tris-HCl. The solution (called sample O⁺) was incubated on ice for 15 min and centrifuged at 42,000 rpm (213,000 × g) using a Ti 50.2 rotor (Beckman) at 4 °C for 2 h. After centrifugation, a thin milky-white top layer (fatty materials) was discarded. Relative to the total retinyl palmitate hydrolysis activity in the original solution, the activities of each fraction were 2% top layer, 82% supernatant, and 16% pellet, and the relative amounts of protein were 1% top layer, 38% supernatant, and 61% pellet, indicating that OSGP efficiently solubilized NREH from crude microsomal membranes and led to a slight increase in enzyme specific activity in the solubilizate. The supernatant (called sample S) was collected and used for DEAE-Sepharose CL-6B ion exchange chromatography.

OSGP-solubilized microsomal supernatant was loaded onto a DEAE-Sepharose CL-6B column previously equilibrated with 10 mM Tris-HCl buffer, pH 7.4 (hereafter referred to as column buffer). Unbound proteins were washed off with column buffer, and the column was then eluted with a gradient of 0-0.3 M NaCl in column buffer. All of the column chromatography procedures were performed in a walk-in cold room unless otherwise specified.

The pooled fractions with retinyl palmitate hydrolase activity from DEAE-Sepharose CL-6B column (designated as sample W) were first treated with ammonium sulfate to a final concentration of 20% (w/v). Then the solution was incubated on ice for 15 min and centrifuged at 12,000 rpm (17,350 × g) in a Ti 50.2 rotor (Beckman) at 4 °C for 15 min. The suspension was collected and loaded onto a Phenyl-Sepharose CL-4B column equilibrated with column buffer containing 20% ammonium sulfate (w/v). The column was eluted as described in the legend to Fig. 1. The detergent/column buffer washing step was performed at room temperature.

Sample WP2, having the highest specific retinyl ester hydrolase activity, was loaded into a Macrosep centrifugal concentrator tube (Filtron Technology Corp., Northborough, MA) with a molecular weight cut-off of 30,000 and centrifuged at 6,000 rpm (4,300 × g) using a SS-34 rotor in Sorvall RC-5B Plus centrifuge (DuPont) at 4 °C for 1 h. No retinyl palmitate hydrolase activity was found in the filtrate. The concentrated protein solution was then loaded onto Sephadex G-100 and Sephacryl S-200 columns connected in series and previously equilibrated with column buffer.

Samples from different purification stages were electrophoresed using the Laemmli discontinuous buffer system (14). To estimate the purity and molecular mass of NREH, 10% SDS-PAGE gels (18 × 16 cm) were run at 60 V for 20 h and stained with silver. The estimated molecular mass of NREH was determined by comparison to the relative mobilities of marker proteins.

Rat liver microsomal carboxylesterase ES-4 and the corresponding polyclonal antibody were prepared as described earlier (15). Polyclonal antibody against mouse serum esterase ES-1 was a kind gift of Dr. Thomas H. Finlay (New York University Medical Center, New York, NY) (16). Generation and characterization of antibodies against peptides corresponding to the carboxyl-terminal ends of ES-4 (12 amino acids) and ES-10 (13 amino acids) will be described in detail elsewhere.² For Western blot analysis, proteins were separated by SDS-PAGE and electrotransferred onto nitrocellulose (Hybond-C, Amersham Corp). The membranes were blocked in 1% bovine serum albumin in Tris/NaCl/Tween 20 and incubated with primary antibodies. The blots were washed in Tris/NaCl/Tween 20; after incubation with peroxidase-conjugated goat anti-rabbit IgG, the membranes were developed with enhanced chemiluminescence (ECL, Amersham Corp.) and exposed to x-ray film.

To analyze the peptide sequence of purified proteins, samples were applied onto a 10% SDS gel. After electrophoresis, the gel was blotted to polyvinylidene difluoride membrane and stained with Amido Black (17). The membrane was sent to the Protein Microsequencing Facility of the Wistar Institute (Philadelphia, PA) for protease digestion and amino acid sequencing.

Protein concentrations were determined by a modified Lowry method (18), using bovine serum albumin as a standard, or assayed by the bicinchoninic acid method (19).

RESULTS

Two rat hepatic microsomal NREHs were purified by a sequence of procedures involving OSGP solubilization, DEAE-Sepharose anion exchange, Phenyl-Sepharose hydrophobic interaction, and Sephadex G-100 and Sepharose S-200 gel filtration chromatographies. OSGP was very efficient in releasing NREH from microsomal membranes (Table I). After anion exchange chromatography of the solubilizate on DEAE-Sepharose, NREH activity was recovered in a single peak in fractions eluted with a 0.0-0.3 M NaCl gradient elution (data not shown). The yield at this stage was 40%, with a 14-fold increase in specific activity over the starting microsomal suspension (Table I). The pooled active fractions were then further separated by Phenyl-Sepharose hydrophobic interaction column chromatography. Two peaks of NREH activity were resolved (Fig. 1). The first broad peak (designated as WP1) eluted with column buffer and a second sharp peak (designated as WP2) was eluted with column buffer containing 1% OSGP. Fraction WP2 showed a 87-fold increase in specific activity over the starting microsomal fraction and showed three bands on SDS-PAGE. The fractions pooled as WP2 were concentrated and loaded onto two columns of Sephadex G-100 and Sephacryl S-200 connected in series. After gel filtration, two peaks of activity peaks were again observed (Fig. 2). The first eluted peak was designated as GS21, and the second peak was designated GS22. Enzyme activity in the solubilizate and the purified enzyme (GS22) was stable when frozen. In contrast, some enzyme activity was lost during the ion exchange and hydrophobic interaction chromatographies. At both of these steps, the activity in pooled and concentrated fractions (shown in Table I) was less than that accounted for by summing the activities in fractions as assayed directly off the columns. Thus it is useful to carry out these steps as quickly as possible.

Table I. Neutral, bile salt-independent retinyl ester hydrolases purification Fraction Volume Protein Total protein Specific activity Yield -Fold ml mg/ml mg units/mg % n Microsomal   Crude (O+) 1087 12.05 13,097 16 100 1 OSGP   Extract (S) 855 5.73 4899 53 123 3 DEAE-Sepharose (W) 831 0.45 370 224 40 14 Phenyl-Sepharose   WP1 405 0.07 27 1050 14 66   WP2 110 0.07 8 1387 6 87 Gel filtration   GS21 22 0.10 3 809 2 50   GS22 36 0.04 3 3078 5 192

The purities of the NREHs were assessed by SDS-PAGE. On a 10% SDS denaturing gel, stained with silver (Fig. 3), sample GS22 appeared as one major band corresponding to a molecular mass of approximately 66 kDa. Sample GS21 contained two major bands corresponding to polypeptides with molecular masses of approximately 58 and 60 kDa. When analyzed by native gel electrophoresis under nondenaturing conditions (data not shown), GS22 migrated as a single band of approximately 70 kDa and the GS21 preparation showed a broad band of high molecular mass (~200 kDa).

Four peptides obtained from the purified enzyme (GS22) were subjected to amino acid sequencing with the following results: peptide 1, FAPPEPAEPWSFVK; peptide 2, ESIPLEFSEDCLYLNIYSP; peptide 3, SFNTVPYIVGFNK; peptide 4, DAGAPTYMYEFQYR. These sequences match exactly those of rat liver carboxylesterase ES-2, as predicted from the nucleotide sequence of the cDNA for this enzyme that was recently determined by one of us (20). In addition, the first eight amino acids of peptide 1 are identical to one of the peptide sequences obtained from purified rat serum esterase (20).

Samples of the two protein bands in GS21 were also subjected to peptide sequencing. Two peptides derived from the lower band (see Fig. 3) had the following sequences: peptide 1, QEFGWIIPTLMGYPLSEGK; peptide 2, TVIGDHGDELFS. These two peptide sequences correspond exactly to sequences found in carboxylesterase ES-10 (also called the pI 6.1 carboxylesterase), as determined from the nucleotide sequence of the cDNA for that enzyme (21). Finally, analysis of three peptides from the upper band in GS21 gave the following sequences: peptide 1, GPLLVQDVVFTDEMAHFDR; peptide 2, FYTEDGNWDLVGNNTPIFFIR; peptide 3, NYFAEVEQMAFD. These sequences correspond exactly to those found in the rat liver catalase monomer (22).

To clarify further the relationships of the preparations containing NREH activity and the major liver carboxylesterases, preparations were electrophoresed on SDS-PAGE, transferred to nitrocellulose, and probed with antibodies to various carboxylesterases. These included mouse carboxylesterases ES-1 (which cross-reacts with rat serum esterase ES-2), ES-4 (also called microsomal hydrolase (MH) or pI 6.4 carboxylesterase), and ES-10 (also called the pI 6.1 carboxylesterase). The anti-pI 6.1 and anti-pI 6.4 are rabbit antipeptide antibodies directed against the COOH-terminal ends of the respective enzymes, whereas the anti-serum esterase antibody is a polyclonal antibody raised against the mouse serum esterase. The results of these studies are shown in Fig. 4. As shown on the left of this figure, one band in the GS21 preparation reacts with the anti-pI 6.1 (ES-10) antibody, whereas GS22 does not react with this antibody. As shown in the center of the figure, neither GS21 nor GS22 react with anti-pI 6.4 antibody, which reacts strongly with the purified pI 6.4 esterase. The right side of the figure shows the pattern of antibody reaction using anti-serum esterase. This antibody is not monospecific and reacts with both purified pI 6.4 enzyme (MH) and with the same band in GS21 as that which reacts with the anti-pI 6.1 antibody. Importantly, however, the antiserum esterase antibody reacts with a protein in rat serum that is the size of authentic serum esterase and with the purified NREH (GS22), which is of intermediate size. An antibody against catalase reacted only with the upper band of GS21 (data not shown).

To characterize purified NREH, a number of experiments were carried out to study the pH dependence, substrate specificity, and enzyme kinetics. The enzyme activity of purified NREH (GS22) functioned from pH 4 to 8 with optimal activity at pH 7.0. The retinyl palmitate hydrolysis activity increased linearly with the incubation time for the first 60 min of the incubation, then gradually slowed down. Initial rates were observed over an enzyme concentration range of 1-6 µg of the purified NREH.

Finally, kinetic parameters for the hydrolysis of retinyl palmitate by purified NREH were determined with various substrate concentrations as shown in Fig. 5. The apparent kinetic parameters were estimated from Lineweaver-Burk plots; K_m and V_max were calculated to be 69 µM and 3.1 nmol/min/mg protein, respectively.

The purified NREH was tested for its substrate specificity to see whether this homogeneous enzyme had other ester hydrolysis activity. Purified NREH showed no hydrolysis activity against cholesteryl oleate. The maximal activity against triolein was about one-tenth that of the hydrolysis of retinyl palmitate under similar assay conditions. This indicates that the purified NREH preferred the retinyl ester as substrate and did not hydrolyze cholesteryl oleate.

DISCUSSION

The studies reported here provide new information on the hepatic enzymes involved in the metabolism of retinyl esters. In addition, they suggest a possible new physiological role for two of the "nonspecific" carboxylesterases of rat liver.

Although the enzymatic hydrolysis of retinyl esters plays a key role in both the hepatic uptake and the mobilization of vitamin A, understanding of the specific enzymes involved is limited. Early work focused on a bile salt-dependent NREH that is now recognized to be carboxylester lipase, a soluble enzyme secreted from the pancreas, liver, and breast of mammals (see Refs. 1 and 7 for reviews). Carboxylester lipase catalyzes the hydrolysis of a wide variety of lipid esters and is more active on cholesteryl esters, triacylglycerols, and lysophospholipids than on retinyl esters.

In addition to carboxylester lipase, the liver and other tissues contain membrane-associated NREH activities that are active in the absence of exogenous bile salts in in vitro assays (4, 9, 10, 23). Evidence suggests that these membrane-associated enzymes may by important in the metabolism of retinyl esters. For example, Boerman and Napoli (23) have shown that apocellular retinol-binding protein directly activates the hydrolysis of endogenous retinyl esters in rat liver microsomes by a bile salt-independent NREH activity. Moreover, chylomicron retinyl esters newly delivered to the liver of intact rats are initially found localized in plasma membrane/endosomal fractions that also contain bile salt-independent NREH activity (11). These latter studies also demonstrated the hydrolysis in vitro of chylomicron retinyl esters by an NREH activity in these membranes.

Despite the potential importance of the bile salt-independent NREHs in retinoid metabolism, little detailed information on their biochemical properties is available. Napoli and colleagues (9) conducted the seminal work in this area by showing that the NREH activity of rat kidney microsomes could be effectively solubilized by the detergent, OSGP. Indeed, our initial attempts at enzyme solubilization built on their work. It is noteworthy that both groups found that the NREH activity could be solubilized in good yield with approximately a 3-4-fold increase in specific activity over the starting microsomes from both tissues. It is also of interest that Napoli et al. (9) were able to resolve two partially purified kidney NREHs using a combination of gel filtration and ion exchange chromatographies. Although their preparations were only slightly enriched over the solubilized microsomes in NREH activity, even at that stage of purification the preparations were more active in catalyzing the hydrolysis of retinyl palmitate than triolein and showed no cholesteryl esterase activity. It seems likely that the preparations studied by these authors contained the carboxylesterases purified and characterized by us as NREHs.

A significant aspect of the work reported here is the isolation in a homogeneous state of a bile salt-independent NREH (GS22) that shows specificity for retinyl esters over triglycerides and that does not catalyze the hydrolysis of cholesteryl esters. This specificity suggests that the enzyme might be important in hepatic retinyl ester metabolism and hence in the body's overall economy of retinoid. The existence of a specific bile salt-independent NREH had been suggested by the work of Napoli et al. (9) discussed above and by our previous studies using active site-directed inhibitors in whole microsomes (10). It is also worth pointing out that the other partially purified NREH isolated in the current studies (GS21) also showed a substrate preference for retinyl esters over triglycerides and did not catalyze the hydrolysis of cholesteryl esters.

A final significant aspect of the work reported here is the demonstration of the probable identity of two bile salt-independent, microsomal NREHs with two of the major carboxylesterases expressed in rat liver. Although the older biochemical literature refers to more than 30 rat liver carboxylesterases, it is now appreciated that almost all of these enzyme activities are manifestations of the gene products of five major loci in linkage group V (24). These esterases are referred to as ES-2 (also called serum esterase), ES-3 (also called the pI 5.6 esterase), ES-4 (also called the pI 6.2/6.4 esterase or microsomal hydrolase), ES-10 (also called the pI 6.0/6.1 esterase), and ES-15 (also called the pI 5.0/5.2 esterase). All of these enzymes have polypeptide monomer molecular masses of 58-65 kDa and all function catalytically as the monomer, except for esterase ES-10, which exists as a homotrimer in the native state (24). All of these enzymes function as carboxylesterases (on a wide variety of oxyester substrates), whereas the microsomal hydrolase, ES-4, also functions as a thioesterase and catalyzes the hydrolysis of long-chain acyl-CoAs (15, 24). The results of our investigations strongly suggest that the two microsomal NREHs isolated here correspond to esterases ES-10 and ES-2.

The evidence suggesting that the partially purified NREH contained in GS21 is esterase ES-10 comes from: 1) the size of the protein as assessed by gel filtration, native gel electrophoresis, and SDS-PAGE; 2) the partial amino acid sequence; and 3) its immunoreactivity. The proteins in GS21 elute early from gel filtration columns and migrate on native gel electrophoresis as a diffuse band with a molecular mass in the 200-kDa range. This is near the known sizes of the homotrimeric esterase ES-10 (180 kDa) and the tetrameric native catalase (240 kDa). Significantly, under denaturing conditions on SDS-PAGE, the preparation shows two distinct bands of molecular masses of approximately 58 and 60 kDa. The 58-kDa species contained amino acid sequences that matched exactly those found in ES-10, whereas the upper band contained sequences that matched exactly the catalase sequence. Finally, antipeptide antibodies directed at the COOH-terminal end of ES-10 specifically recognized the 58-kDa protein in GS21.

Taken together, the results suggest that ES-10 (the pI 6.0/6.1 esterase) can function as a bile salt-independent retinyl ester hydrolase. Our results agree quite well with the studies of Mentlein and Heymann (25). These authors studied the hydrolysis of retinyl palmitate in vitro by four purified, rat liver carboxylesterases, namely ES-3, ES-10, and the two forms (pI 6.2 and 6.4) of ES-4. Using a variety of substrate presentation forms (all of which were different from the assay used by us), they found that ES-4 enzyme showed the highest REH activity and that ES-10 showed lower activity (<10% of ES-4) under some conditions of substrate presentation. Our results obviously do not address the possibility that ES-4 is a NREH, because it is possible that this enzyme was present in the first peak of NREH activity that eluted from Phenyl-Sepharose (WP1), which we did not further characterize. It is clear, however, based on our Western blotting experiments, that neither preparation isolated by us contains ES-4. It appears that both ES-4 and ES-10 can function as NREHs. It is important to point out, however, that both of these enzymes are localized in endoplasmic reticulum-derived vesicles in the microsomal fraction, and both we (4, 10, 11) and Mentlein and Heymann (25) agree that most of the membrane-associated NREH activity in rat liver membranes is associated with plasma membrane/endosomal vesicles. An enzyme more likely to be localized in such vesicles might be the ES-2 or a related enzyme as discussed below.

The highest NREH specific activity was associated with the nearly homogeneous protein in GS22. Evidence that this protein is identical to or highly related to the serum esterase ES-2 comes from two sources. First, the isolated NREH reacted with antibodies to the purified serum esterase (20). Although this evidence is suggestive of a high degree of relatedness of the NREH and ES-2, it does not prove that the two proteins are identical. This is especially true given that the anti-serum esterase antibodies used for the Western blots cross react with other related but distinct carboxylesterases such as ES-4 and ES-10 (Fig. 4). Although, in addition (as also shown in Fig. 4), the purified NREH runs slightly faster on SDS-PAGE than does the mature esterase in rat serum, this is probably because of the isolation of the serum esterase from intracellular sources in a precursor form, which undergoes further glycosylation before being secreted (26). Second, there was complete identity of the amino acid sequences of four internal peptides (60 residues total) with sequences in ES-2, strongly suggesting that the two proteins are identical. If the NREH is identical to serum esterase, the enzyme may function on chylomicron retinyl esters at or near the cell surface in the space of Disse, as has been suggested for the metabolism of chylomicron remnants by other secreted lipases such as hepatic lipase and carboxylester lipase. If the NREH is a distinct enzyme (or a retained form of ES-2), it may be the plasma membrane/endosomal NREH studied previously in whole membrane preparations and thought to be important in hepatic retinyl ester metabolism.