Membrane Topology and Cell Surface Targeting of Microsomal Epoxide Hydrolase

Microsomal epoxide hydrolase (mEH) is a bifunctional membrane protein that plays a central role in the metabolism of xenobiotics and in the hepatocyte uptake of bile acids. Numerous studies have established that this protein is expressed both in the endoplasmic reticulum and at the sinusoidal plasma membrane. Preliminary evidence has suggested that mEH is expressed in the endoplasmic reticulum (ER) membrane with two distinct topological orientations. To further characterize the membrane topology and targeting of this protein, an N-glycosylation site was engineered into mEH to serve as a topological probe for the elucidation of the cellular location of mEH domains. The cDNAs for mEH and this mEH derivative (mEHg) were then expressed in vitro and in COS-7 cells. Analysis of total expressed protein in these systems indicated that mEHg was largely unglycosylated, suggesting that expression in the ER was primarily of a type I orientation (Ccyt/Nexo). However, analysis, by biotin/avidin labeling procedures, of mEHg expressed at the surface of transfected COS-7 cells, showed it to be fully glycosylated, indicating that the topological form targeted to this site originally had a type II orientation (Cexo/Ncyt) in the ER. The surface expression of mEH was also confirmed by confocal fluorescence scanning microscopy. The sensitivity of mEH topology to the charge at the N-terminal domain was demonstrated by altering the net charge over a range of 0 to +3. The introduction of one positive charge led to a significant inversion in mEH topology based on glycosylation site analysis. A truncated form of mEH lacking the N-terminal hydrophobic transmembrane domain was also detected on the extracellular surface of transfected COS-7 cells, demonstrating the existence of at least one additional transmembrane segment. These results suggest that mEH may be integrated into the membrane with multiple transmembrane domains and is inserted into the ER membrane with two topological orientations, one of which is targeted to the plasma membrane where it mediates bile acid transport.

Microsomal epoxide hydrolase (mEH) is a bifunctional membrane protein that plays a central role in the metabolism of xenobiotics and in the hepatocyte uptake of bile acids. Numerous studies have established that this protein is expressed both in the endoplasmic reticulum and at the sinusoidal plasma membrane. Preliminary evidence has suggested that mEH is expressed in the endoplasmic reticulum (ER) membrane with two distinct topological orientations. To further characterize the membrane topology and targeting of this protein, an N-glycosylation site was engineered into mEH to serve as a topological probe for the elucidation of the cellular location of mEH domains. The cDNAs for mEH and this mEH derivative (mEHg) were then expressed in vitro and in COS-7 cells. Analysis of total expressed protein in these systems indicated that mEHg was largely unglycosylated, suggesting that expression in the ER was primarily of a type I orientation (Ccyt/Nexo). However, analysis, by biotin/avidin labeling procedures, of mEHg expressed at the surface of transfected COS-7 cells, showed it to be fully glycosylated, indicating that the topological form targeted to this site originally had a type II orientation (Cexo/Ncyt) in the ER. The surface expression of mEH was also confirmed by confocal fluorescence scanning microscopy. The sensitivity of mEH topology to the charge at the N-terminal domain was demonstrated by altering the net charge over a range of 0 to ؉3. The introduction of one positive charge led to a significant inversion in mEH topology based on glycosylation site analysis. A truncated form of mEH lacking the N-terminal hydrophobic transmembrane domain was also detected on the extracellular surface of transfected COS-7 cells, demonstrating the existence of at least one additional transmembrane segment. These results suggest that mEH may be integrated into the membrane with multiple transmembrane domains and is inserted into the ER membrane with two topological orientations, one of which is targeted to the plasma membrane where it mediates bile acid transport.

(EC 3.3.2.3)
is expressed in the endoplasmic reticulum (ER) membrane, where it is involved in the metabolism of many xenobiotics such as polycyclic aromatic hydrocarbon carcinogens (1), in concert with other proteins such as members of the cytochrome P450 enzyme superfamily (2,3), dihydrodiol dehydrogenase (4), and glutathione S-transferase (5). Numerous studies (6 -12) have established that this protein is also expressed at the plasma membrane where it is able to mediate the sodium-dependent uptake of bile acids. The bile acids play an important role in many physiological processes such as (a) digestion; (b) the formation of bile, which functions as an excretory vehicle for numerous compounds such as cholesterol and metabolites of drugs and carcinogens; (c) the regulation of cholesterol metabolism; and (d) the modulation of hepatocyte signaling pathways. The surface location of mEH was initially suggested by enzyme marker analysis (13) and further confirmed using monoclonal antibodies against mEH (8), cDNA transfection procedures (12), and confocal immunofluorescence microscopy of hepatocytes and transfected Madin-Darby canine kidney cells. 2 In addition to mEH, several members of the cytochrome P450 superfamily and NADPH cytochrome P450 reductase are also expressed on both the cell surface and in the ER membrane (14 -17). Epitope mapping and transport studies have demonstrated that mEH is, in fact, expressed in the ER with two distinct topological orientations derived from a single population of nascent chains (18). The factors that lead to the formation of multiple topologies and the targeting of this bifunctional protein to multiple sites have, however, not been fully elaborated. To further characterize the membrane organization and topological forms of mEH in the ER and plasma membrane, we have used glycosylation site probe insertion, biotin/avidin labeling procedures, and confocal fluorescence microscopy, techniques that have been used extensively to study membrane protein topology and targeting (15, 19 -24). These studies have established that mEH is integrated into the membrane with multiple transmembrane domains and is expressed in the ER with a type I (Ccyt/Nexo) and type II (Cexo/Ncyt) topological orientation, with the type II form targeted to the plasma membrane.

EXPERIMENTAL PROCEDURES
Cell Culture and Transfection Procedures-COS-7 cells were obtained from American Type Culture Collection and were transfected with rat cDNA for mEH or mEH derivatives in the pcDNA1.1/Amp vector (Invitrogen) using the DEAE-dextran procedure to obtain transient transformants (25). Cells were grown at 37°C in Dulbecco's modified Eagle's medium supplemented with 10% fetal calf serum. mEH cDNA/pBSK Construct-The cDNA for mEH was prepared from rat liver poly(A) ϩ mRNA using reverse transcription and the polymerase chain reaction procedure (26). The 5Ј primer used, ATGC-GTCGACGGCACTTCTGTTCCCAGGGAAA, was targeted to the cDNA sequence 85-106 and contained an SalI site (underlined). The 3Ј antisense primer used, AAGCGGTACCCTTCCATAAGTGTGACTTGGGA-GG, was targeted to nucleotide 1582-1559 and contained a KpnI site (underlined). The polymerase chain reaction product was digested with SalI and KpnI and cloned into the vector pRSV. The mEH cDNA was then isolated as an SalI/EcoRI (linker site) fragment and inserted into pBSK (Bluescript/Stratagene) digested with SalI and EcoRI. The fidelity of the cDNA was verified by restriction enzyme mapping and by DNA sequencing using the dideoxy termination method (27).
Preparation of the mEH/P-glycoprotein Fragment Chimeric cDNA Construct (mEH/Pgpf)-In this construct, a Pgp cDNA PstI(nt 1314)/ HindII(nt 2033) fragment containing amino acid residues 394 -678 (28) was attached to the 3Ј end of the mEH coding sequence through the PstI site at the mEH cDNA nt 1513, in a three-way ligation: mEH XhoI-(linker)/PstI fragment, plus the Pgp PstI/HindIII fragment, and the mEH/pBSK construct in which the XhoI/HindIII fragment had been removed as shown in Fig. 1A. The resulting construct encoded the entire mEH polypeptide (455 amino acids), followed by a Pgp polypeptide tail (Pgp amino acid 439 to 678), which contained a glycosylation site at position 491, followed by the last 15 amino acids (440 to 455) of mEH. These terminal 15 amino acids thus appeared twice in the chimeric protein in both the mEH portion and again at the C terminus of the chimeric protein. The mEH termination codon following the last 15 amino acids was used to terminate the translation of the chimeric polypeptide. To modify the net charge at the N terminus of the chimeric cDNA, the XhoI/EcoRV(983) fragment of the chimeric construct was replaced with a XhoI/EcoRV fragment isolated from the mEHg/(ϩ3) cDNA, which contained the engineered positive charge additions as described below.
In Vitro Transcription/Translation of mEH and mEH Derivatives-The expression of mEH and mEH derivatives was investigated by in vitro transcription/translation with the various mEH cDNA constructs inserted in pBSK. These constructs were linearized by EcoRI digestion and in vitro transcribed with T 3 RNA polymerase (Ambion/MEGAscript T3 kit). The resultant mEH mRNA (1 g) was then translated in vitro with the rabbit reticulocyte lysate system (Promega) using conditions recommended by the manufacturer in the presence and absence of canine microsomal membranes (Promega). [ 35 S]methionine (40 Ci) was included to radiolabel newly synthesized protein. The effect of the glycosylation inhibitor tripeptide Asn-Tyr-Thr (NYT) on the synthesis of mEH derivatives containing an introduced glycosylation site was evaluated as described previously (29). The translated products were analyzed by sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) and visualized by autoradiography.
Addition of an Internal Glycosylation Site-A glycosylation site (Asn-Gly-Thr) was introduced near the C terminus of mEH (mEHg) by replacing the wild type Phe-Ala-Ala (residue 432-434, Fig. 2). The mutations were made by replacing the wild type SphI/HindIII fragment (nt 1370 to 1463) with a synthesized mutant SphI/HindIII fragment. To prepare this mutant fragment, two oligonucleotides were synthesized: TCCGAGCTACTGCATGCCCCAGAAAAGTGGGTGAAGGTCAAGTA-CCCCAAACTCATCTCCTATTCCTAC (oligonucleotide f in Fig. 1B in sense sequence, nucleotide 1354 to 1423; the SphI site is underlined) and CTGGGCCAGAAGCTTGGGCTCTTCAAAGGTACCATTGTGGC-CCCCACGTTCCATGTAGGAATAGGAGAT (oligonucleotide g in Fig. 1B in antisense sequence, nucleotide 1477 to 1408; mutated nucleotides are in bold face. The HindIII site AAGCTT and the mutation-created site KpnI GGATCC are underlined). These two oligonucleotides were annealed to each other through the complementary 15 nucleotides (shown in italics) and were extended to full-length double-stranded DNA with Klenow large fragment of DNA polymerase 1. The extended DNA fragment was digested with restriction enzymes SphI and HindIII and inserted into a mEH/pBSK construct in which the wild type SphI/ HindII fragment had been removed. The authenticity of the construct (mEHg/pBSK) was confirmed by restriction enzyme mapping (the intactness of the SphI and HindIII sites, the appearance of the new KpnI site and the length of the mutated SphI/HindIII fragment as compared with the wild type) and by DNA sequencing. For transfection into COS-7 cells, the mutant cDNA insert was isolated from the pBSK construct as a XhoI/XbaI fragment and inserted into the pcDNA1.1/ Amp vector (Invitrogen) digested with XhoI and XbaI.
Charge Modifications at the N terminus of mEH-The cDNA sequence changes for charge alterations at the N terminus were performed with a similar annealing-extension-replacement procedure as was used for engineering the glycosylation site. Four different mutant oligonucleotides in sense sequence (Fig. 1B, a-d, corresponding to cDNA nucleotide 124 to 180 in position) were separately annealed to a common antisense oligonucleotide e (nucleotide 223 to 165) through the complementary 15-nucleotide sequence and extended into full-length double-stranded DNA fragments by Klenow large fragment. These fragments were then digested with PstI and SmaI and inserted into wild type mEH/pBSK construct in which the PstI/SmaI fragment had been removed. To add the glycosylation site to these charge-modified mEH cDNAs, the wild type SalI (at the 5Ј end of the cDNA)/EcoRV(nt 983) fragment in the mEHg/pBSK construct was replaced by the mutant SalI/EcoRV fragments isolated from the four charge-modified pBSK constructs. The Glu to Gly substitution at position 4 was effected with oligonucleotides a and e (

SEQUENCE 1
This DNA fragment was formed by annealing two oligonucleotides together (Fig. 1B, oligonucleotides h and i). It contained a PstI sticky end on the left and a SmaI blunt half-site on the right and an ATG translation start codon (bold phase). The codon TCC corresponded to the Ser codon at residue 20 of the wild type mEH polypeptide. Thus, this replacement links the ATG codon directly to residue 20, deleting residues 2 to 19. This truncated cDNA is designated mEHt. Confirmation of the sequence change and transfer of the mutant cDNA insert from the pBSK construct to pcDNA1.1/Amp vector was performed as described above.
Western Blot Analysis-Western analyses were performed using a polyclonal antibody against mEH and the ProtoBlot alkaline phosphatase detection system (Promega).
Cell Surface Biotinylation-The expression of mEH and mEH derivatives on the plasma membrane of COS-7 cells was established using a cell surface biotinylation labeling procedure with sulfosuccinimidyl-6-(biotinamido)hexanoate (Sulfo-NHS-Biotin) (Pierce) on transfected and untransfected cells as described previously (17). The labeled surface proteins were isolated using avidin-agarose beads, and mEH and the mEH derivatives were detected by SDS-PAGE and Western blot analysis. To establish that the reagent labeled only cell surface protein, the biotinylation of the cytoplasmic protein, Hsp 70 was also determined using an anti-Hsp 72/23 monoclonal antibody (Roche Molecular Biochemicals) as described previously (21). In addition, COS-7 cells were disrupted by suspending in 20 mM Tris, pH 7.4 (1 ml), frozen and thawed (2ϫ) and homogenized with 100 strokes in a tight Dounce homogenizer. The soluble fraction isolated by centrifugation was biotinylated as described for intact cells. Following biotinylation, the excess reagent was removed by dialysis against phosphate-buffered saline for 24 h, and the resultant protein was treated as described for intact cells and the biotinylated Hsp 70 determined as described above.
Deglycosylation of mEH Derivatives-Glycosylated forms of mEH derived from the COS-7 cell plasma membrane or directly from the microsome fraction were deglycosylated with Endo H f (New England Biolabs) using procedures supplied by the manufacturer. Proteins were then detected by SDS-PAGE and Western blot analysis.
Inhibition of the Glycosylation Reaction by the Tripeptide NYT-The tripeptide Asn-Tyr-Thr, with both the N and C termini blocked as acetyl-Asn-Tyr-Thr-amide, has been shown to be an effective substrate for oligosaccharyl transferase (29) and to inhibit glycosylation reactions in vitro (30). This tripeptide was kindly provided by Dr. R. A. F. Reithmeier (University of Toronto, Canada) and was included in the in vitro translation mixture at a final concentration of 46 M to inhibit the in vitro glycosylation reaction of the mEHg(ϩ3) derivative.
Proteinase K Digestion-After in vitro translation, proteinase K was added to a final concentration of 0.1 mg/ml, followed by incubation at 0°C for 60 min as recommended by Roche Molecular Biochemicals. Phenylmethylsulfonyl fluoride was then added to a final concentration of 4 mM to stop the digestion before the reaction product was analyzed with SDS-PAGE.
Confocal Immunofluorescence Analysis of Transfected COS-7 Cells-COS-7 cells transfected with the cDNA for mEH, mEHg(0), mEHg(ϩ3), and mEHt were incubated with anti-mEH monoclonal antibody, mAb 25D-1 (1 g/ml) for 18 h at 4°C, and fixed with 4% paraformaldehyde before incubation with a secondary antibody, cy3-conjugated antimouse IgG for 4 h at 24°C. Antibody labeling was characterized by cy3 epifluorescence using a Zeiss LSM-210 scanning confocal microscope equipped with a barrier filter. Image analysis was performed using the standard system operating software.

Addition of an N-Glycosylation Site and N-terminal Charge
Modification of mEH-The membrane topology of mEH was characterized by the introduction of N-glycosylation sites into two different loci of mEH, which itself lacks such sites, to establish whether a particular protein domain is expressed in the lumen of the endoplasmic reticulum. A fragment of Pglycoprotein (Pgpf) (amino acids 394 -678) containing a potential glycosylation site at amino acid 439 (28), which has been used to study the topology of the cystic fibrosis transmembrane conductance regulator (19), has been linked to mEH at position 454 as shown in Fig. 1A. A glycosylation site was also directly introduced into full-length mEH at position 432 by site-directed mutagenesis (Fig. 1B) resulting in the conversion of Phe-Ala-Ala to Asn-Gly-Thr (Fig. 2). In addition, the net charge of the N-terminal end preceding the 16-residue hydrophobic domain was modified over a range of 0 to ϩ3 as shown in Fig.  2 to investigate the role of charge in regulating mEH insertion into the ER membrane in vitro and COS cell expression systems, and also to define the topological form of mEH that is targeted to the plasma membrane.
In Vitro Translation of mEH and mEH Derivatives-The various mEH constructs in pBSK were in vitro transcribed with T 3 RNA polymerase and then translated with the rabbit reticulocyte lysate system (Promega) in the presence and absence of canine pancreas microsomes. The mEH(0)/Pgpf chimera containing a glycosylation site was translated yielding a peptide with the expected molecular mass (80.7 kDa). However, in the presence of microsomes no glycosylated higher molecular weight species was observed indicating that an undetectable amount of the Pgpf tail on mEH was located in the ER lumen, thus suggesting primarily a type I topology in the ER (Fig. 3A,  lanes 1 and 2). A mEH derivative in which Trp (position 2) was converted to Arg, and Glu (position 4) was converted to Lys, which results in the alteration of the net N-terminal charge from 0 to ϩ3 (Fig. 2), was then coupled to Pgpf and expressed in the cell-free system. In the absence of microsomes, a single band was observed that had the same mobility as the chimera composed of mEH(0)/Pgpf (Fig. 3A, lane 3). In the presence of microsomes, however, two bands were observed, indicating that approximately 40% of the expressed protein was glycosylated, and suggesting that the glycosylation site must be partially located in the lumen (Fig. 3A, lane 4). Full-length mEH was also expressed in vitro and shown by SDS-PAGE/autoradiography to have the correct molecular weight (Fig. 3B, lane 1) and to react with an anti-mEH antibody (25D-1) (data not shown). When mEH with a glycosylation site introduced at position 432 (mEHg(0)) was expressed in the in vitro system, only one band was observed in the presence or absence of microsomes, with a mobility identical to unmodified mEH (Fig. 3B, lanes 2 and 3) indicating that the glycosylation site was located outside of the ER lumen and that the protein primarily adopted a type I orientation. When the mEHg(ϩ3) derivative (Fig. 2) was expressed in the absence of microsome, a single band was observed (Fig. 3B, lane 4). However, as observed for the mEH(ϩ3)/Pgpf derivative, in the presence of microsomes, two bands were formed (Fig. 3B, lane 5) suggesting that approximately 55% of the translated protein was oriented with the glycosylation site in the ER lumen. This conclusion was supported by the observation that the formation of the upper band was significantly inhibited (Fig. 3B, lane 6) by the glycosylation tripeptide inhibitor Asn-Tyr-Thr (30). In addition, mEHg(ϩ3) was largely protected from proteinase K digestion, whereas the product formed in the absence of microsomes was degraded (Fig. 3B, lanes 7 and 8), further supporting the luminal orientation of the glycosylation site at position 432.
Expression of mEH and mEH Derivatives in Transfected COS-7 Cells-The mEH constructs in pcDNA1.1/Amp were transiently expressed in COS-7 cells to corroborate the results from the cell free system and extend the search for evidence of multiple topologies of mEH, which have been described in previous studies (18). Total lysates from cells transfected with mEHg(0) afforded primarily a single band (Fig. 4, lane 2) with a mobility identical to unmodified mEH (Fig. 4, lane 1). The presence of a trace amount of a protein product with a lower mobility (lane 2) is also detected, demonstrating the existence of a higher molecular weight glycosylated species resulting from an alternate topology with the glycosylation site located in the ER lumen. In contrast, the expression of mEHg(ϩ3) afforded a single band with an increased molecular weight (Fig.  4, lane 3), demonstrating that this derivative had been fully glycosylated as a result of a type II orientation with the glycosylation site (position 432) expressed entirely in the ER lumen.
The partial glycosylation of this derivative observed in vitro indicated that the cell free system was not as efficient as the COS-7 cell system in expressing the type II orientation. Glycosylation of the mEHg(ϩ3) product was confirmed by deglycosylation with Endo H f , resulting in a single band with the mobility of mEH (Fig. 4, lane 4).
The expression of mEH and mEHg(0) at the cell surface of transfected COS-7 cells was next investigated using biotinylation labeling procedures as previously described for aromatase P450 (17) and the ileal sodium-dependent bile acid transporter (21), in an effort to obtain definitive evidence that mEHg(0) as well as mEH with a type II topology in the ER membrane, was targeted to the plasma membrane. As shown in Fig. 5 (lane 1), the biotinylated cell surface product obtained from the COS-7 cells expressing mEH without an inserted glycosylation site, afforded a single band with a mobility identical to mEH obtained from the total cell lysate (Fig. 4, lane 1), suggesting that some percentage of the expressed protein was accessible to the labeling reagent on the cell surface. In contrast, the surface biotinylation of COS-7 cells expressing mEHg(0) afforded only a protein with a higher molecular weight (Fig. 5, lane 2) than the main protein found in the total cell lysate (Fig. 4, lane 2). This material was also treated with Endo H f as described above for mEHg(ϩ3) (Fig. 4, lane 4), yielding a product with the same molecular weight as mEH (Fig. 5, lane 3), thereby establishing that the increased size of this product resulted from glycosylation of the introduced glycosylation site in the ER lumen. This result confirmed the thesis that mEH can exist in two topological forms in the ER with the type II form targeted to the plasma membrane. mEHg(ϩ3), which afforded only a fully glycosylated product (Fig. 4, lane 3) in the ER, was also targeted to the plasma membrane, where biotinylation identified the glycosylated derivative (Fig. 5, lane 4). It was established that the biotinylating reagent reacted only with surface proteins by demonstrating that unglycosylated mEHg(0) was not labeled in intact cells (Fig. 5, lane 2), although it is the major intracellular form of the protein (Fig. 4, lane 2). This conclusion was confirmed by demonstrating that the endogenous cytoplasmic protein, Hsp 70 was labeled only in cell lysates and not in intact cells (data not shown).
The Effect of N-terminal Charge on mEH Topology-The alteration of the net N-terminal charge from 0 to ϩ3 has been shown to result in a change in mEH topology from primarily a type I to a type II orientation. To further evaluate the role of charge on the topological orientation of mEH, three additional derivatives were prepared with altered N-terminal charges at different positions. When the negative charge at position 4 was removed (Glu to Gly), the resultant derivative (mEHg(ϩ1)a) (Fig. 2) was, like mEHg(0), primarily expressed in the ER with a type I orientation (Fig. 4, lane 5) as estimated by the level of glycosylation (Table I). When a positive charge was added at   4. Expression of mEH and mEH derivatives in COS-7 cells. mEH and mEH derivatives were transiently expressed in COS-7 cells, and the total cell lysate analyzed by SDS-PAGE and immunoblotting using a polyclonal antibody against mEH and the ProtoBlot alkaline phosphatase detection system (Promega). 100 g of protein were applied to lanes 1-8. mEHg derivatives have a glycosylation site at position 431 with N-terminal net charges of 0, ϩ1a, ϩ1b, ϩ2, and ϩ3. mEHt indicates truncated mEH (ϪN-terminal 19 amino acids). mEHg(ϩ3)d indicates deglycosylated mEHg(ϩ3). position 2 (Trp to Arg), the resultant derivative (mEHg(ϩ1)b) had the same net N-terminal charge as mEHg(ϩ1)a but with a different charge display, was almost totally converted to the inverted topology (type II) (Fig. 4, lane 6). When the negative charge at position 4 was converted to a positive charge (Glu to Lys), the resultant derivative, (mEHg(ϩ2)), with one additional positive N-terminal charge, was now expressed in the ER with approximately 55% in the type II orientation (Table I; Fig. 4,  lane 7). The introduction of an additional N-terminal positive charge (mEHg(ϩ3)) afforded only type II orientation as described above (Fig. 4, lane 3). These results suggest that the final topological orientation of mEH is exquisitely sensitive to factors such as net N-terminal charge and charge location.
Expression of an N-terminal Truncated (2)(3)(4)(5)(6)(7)(8)(9)(10)(11)(12)(13)(14)(15)(16)(17)(18)(19) Derivative of mEH-A N-terminal truncated derivative of mEH (mEHt), which is missing amino acids 2-19 was constructed to further characterize the number of transmembrane domains in mEH. Hydropathy analysis has suggested the presence of 4 (11) or 6 (31) domains that integrate into the membrane. Other studies, however, have suggested the existence of only a single transmembrane domain comprising amino acids 4 -20 (32,33). The cDNA of mEHt was transfected into COS-7 cells, and analysis of total protein indicated the synthesis of a protein at approximately the same level of expression as the other derivatives in this study, with the predicted decrease in molecular weight (Fig. 4, lane 8). Biotinylation of COS-7 cells expressing this derivative resulted in the identification of this protein product on the cell surface (Fig. 5, lane 5), which would only be possible if the truncated protein integrated into the ER membrane before expression at the plasma membrane. These results, therefore, suggest the presence of at least one transmembrane domain in addition to the N-terminal hydrophobic domain.
Confocal Immunofluorescence Analysis-The expression of mEHg(0) on the surface of transfected COS-7 cells was further established using confocal fluorescence microscopy, which detected the specific interaction of an anti-mEH monoclonal an-tibody (25D-1) with an mEH epitope on the cell surface as shown in Fig. 6A. Similar results were obtain with mEH, mEHg(ϩ3), and mEHt (Fig. 6, B-D). These studies establish the existence of the luminal orientation of a mEH epitope before targeting to the plasma membrane and are consistent with the glycosylation studies. No fluorescent rings were observed in the absence of transfection or when nonimmune IgG was used instead of the primary antibody (Fig. 6, E and F). Permeabilized cells transfected with mEHg(0) showed fluorescence throughout the body of the cell (Fig. 6G). DISCUSSION Previous studies have demonstrated that mEH is targeted to both the ER and plasma membrane, where it mediates the transport of bile acids (8,11,12,18). Epitope accessibility and resistance to proteolysis suggested that mEH was expressed in the ER membrane with two distinct topological orientations (18), because a common epitope for mEH was found on both the hepatocyte cell surface and on the cytoplasmic face of the ER. To further establish the topology and targeting properties of mEH, use has been made of glycosylation site insertion, biotin/ avidin labeling technology and confocal fluorescence scanning microscopy. Initial studies using an in vitro expression system with a glycosylation site located either in a P-glycoprotein fragment linked to mEH (mEH/Pgpf) or directly engineered into mEH at position 432 (mEHg(0)) indicated that a single unglycosylated product was formed (Fig. 3A). Alteration of the N-terminal charge from 0 to ϩ3 (mEH[ϩ3]), however, afforded a glycosylated product, establishing that this alteration in charge led to a topological inversion that placed the glycosylation site in the ER lumen. The addition of the internal glycosylation site at position 432 was affected with conservative substitutions (Fig. 2) and located far from the N-terminal transmembrane domain in a region of mEH that exhibits numerous amino acid variations between species (31,34,35) where total protein expression levels were approximately the same as observed for unmodified mEH.
The expression of mEHg(0) was also carried out in an intact COS-7 cell system where, in a total cell lysate, mEHg(0) afforded primarily an unglycosylated form, with a small percent of the higher molecular weight glycosylated protein (Fig. 4,  lane 2), whereas the modified mEHg(ϩ3) was expressed only in the glycosylated form (Fig. 4, lane 3). These glycosylation results are qualitatively similar to those previously reported for mEH when the potential glycosylation site was inserted at position 39 or 303 (20). These studies, however, analyzed only total expressed protein with an indicated lower limit of detec- FIG. 5. Analysis of cell surface expression of mEH and mEH derivatives in transfected COS-7 cells by biotinylation procedures. mEH and mEH derivatives were transiently expressed in COS-7 cells. Cell surface proteins were labeled with succinimidyl-6-(biotinamido)hexanoate and isolated using avidin-agarose. Bound proteins were then analyzed by SDS-PAGE and immunoblotting as described in Fig. 3. Biotinylated surface protein from 2 to 15 100-mm plates was applied to each lane. mEHg(0)d indicates deglycosylated mEHg(0).

TABLE I
The effect of N-terminal net charge and position on the topological orientation of mEHg expressed in transfected COS-7 cells mEHg derivatives with different N-terminal net charge (0 to ϩ3) were expressed in COS-7 cells. The percentage of type II (Ncyt/Cexo) orientation was established by the degree of observed glycosylation as described in Fig. 4.
a Refers to the N-terminal 4 amino acids as described in Fig. 2. tion of 5-10% (36) and did not investigate mEH expressed at the plasma membrane. Analysis of intact transfected COS-7 cells with the biotin/avidin surface labeling procedure established that mEHg(0) that was targeted to the plasma membrane was in the glycosylated form (Fig. 5, lane 2), thereby establishing a luminal type II orientation for a fraction of the total expressed protein. Native mEH was also biotinylated on the cell surface, but the protein targeted to the plasma membrane had an unaltered molecular weight, as expected in the absence of a glycosylation site. The presence of mEHg(0) (Fig.  6) as well as mEH, mEHg(ϩ3), and mEHt on the surface of transfected COS-7 cells was corroborated using confocal fluorescence spectroscopy, confirming that a portion of the 25D-1 epitope was originally in the ER lumen, because it reacts with the antibody on the cell surface. Similar results were observed for Madin-Darby canine kidney cells stably transfected with mEH and for intact hepatocytes. 2 These results are consistent with our earlier studies, which showed that an anti-mEH monoclonal antibody (25A-3) could protect both hepatocytes (8) and transfected Madin-Darby canine kidney cells expressing mEH (12) from DIDS inhibition of taurocholate transport, and which demonstrates that these cells possess a percentage of the type II form of mEH, which would result in expression of mEH on the extracellular surface where it could mediate the binding and transport of bile acids. The relationship between total mEH expression and cell surface expression is under investigation.
Various factors have been shown to influence the topological orientation of proteins in the ER membrane, such as the distribution of charged residues flanking the signal anchor sequence, where positive charges are enriched on the cytosolic side and depleted from the luminal (exoplasmic) side of the first transmembrane domain (37)(38)(39). The role of charge on membrane protein topology has been confirmed by site-directed mutagenesis (40). Protein topology is also influenced by the folding properties of the N-terminal sequence (41) as well as the length and hydrophobicity of the transmembrane segment (39,42,43). The resultant topology (or topologies) of a membrane protein is thus determined by a complex interaction of these factors. The charge distribution (⌬[C-N]) around the Nterminal anchor of mEH (Ϫ2 for rat, see Ref. 31; Ϫ3 for human, see Ref. 34; and Ϫ4 for rabbit, see Ref. 35) predicts a type II orientation. However, a majority of mEH in the ER is found in the opposite type I orientation, stressing that multiple factors acting in concert determine protein topology, and can lead, in some cases, to the expression of more than one topological form. The effect of charge alterations reported in this study (Table I) stresses the exquisite sensitivity of mEH topology to this variable so that subtle variations have a dramatic effect on the ratio of the two topological forms. Studies with aromatase cytochrome P450 offer a clear illustration of the above paradigm. Based on the degree of glycosylation of a naturally occurring glycosylation site, this protein is expressed in two topological orientations (17), with one of the two forms expressed at the plasma membrane. Several other membrane proteins have been described, which are expressed in more than one topological form, such as the prion protein (44), P-glycoprotein (45), and ductin (46).
Sequence motifs known to act as retrieval signals for resident ER proteins with a Nexo (47) or Ncyt (48) orientation are not found in mEH. Structural features within the N-terminal hydrophobic domain may also function as an ER retention signal (49) as described for cytochrome P450 2C1, perhaps by mediating the formation of homo-or hetero-oligomers, which prevent transport from the ER compartment. Other proteins destined to reside in the plasma membrane leave the ER by default and travel along the exocytotic pathway (50). The two topological forms of mEH may undergo differential oligomerization, resulting in the expression of mEH in the plasma membrane by default or through a specific targeting pathway. This idea is again illustrated by studies with aromatase cytochrome P450 that have established that only one of the two topological forms present in the ER membrane is targeted to the plasma membrane (17). Several other members of the cytochrome P450 superfamily together with NADPH:cytochrome P450 reductase (17) have also been shown to be expressed on the hepatocyte cell surface (51) and in the ER. The two topological forms of ductin described above are targeted to different cellular domains where this protein functions as either a component of a connexon channel of gap junctions or as subunit c of the vacuolar H ϩ -ATPase (46).
The number of transmembrane domains that anchor mEH in the ER has been the subject of numerous studies. Hydropathy analysis indicates the presence of a hydrophobic N-terminal region as well as several additional domains that could integrate into the membrane (11,31). Studies using truncated mEH suggested that mEH has only a single transmembrane domain, based on the results of alkaline extraction of membranes containing mEH or the truncated derivative (32). The expression of this truncated derivative, however, was only 5% of the level observed for mEH in BHK cells, and the authors failed to point out that approximately 40% of the truncated mEH appeared to be resistant to this extraction procedure, thereby leaving the conclusion of this study open to question. The identification of a truncated mEH derivative (mEHt) on the surface of transfected COS-7 cells (Fig. 5, lane 5) clearly establishes that mEH must integrate into the membrane with more than one transmembrane domain to have mEHt targeted and expressed on the extracellular surface. Similar conclusions have been obtained with aromatase P450, where a truncated version of this protein was also identified on the surface of transfected COS-7 cells (17), suggesting the presence of at least one additional transmembrane domain apart from the N-terminal hydrophobic domain. These results are supported by previous studies on cytochrome P450 IIE1 (52) and P450 1A1 (53). Because mEH residues 39, 303 (20), and 432 (this study) are expressed on the same side of the membrane in either the type I or type II orientation, based on glycosylation site analysis, a model with three transmembrane domains would be consistent with the current available data. The oligomerization of mEH could then lead to a system spanning the membrane multiple times, which would be consistent with its role as a bile acid transporter. Evidence for mEH oligomerization with itself as well as with cytochrome P450 and NADPH:cytochrome P450 reductase has been reported (54).
In conclusion, these studies have demonstrated that mEH is expressed with two distinct topological orientations in the ER membrane and that orientation is extremely sensitive to alterations in charge in the N-terminal 5 amino acids. The use of glycosylation site insertion and surface biotinylation has also established that one of these topological forms is targeted to the plasma membrane where it can function as a bile acid transport protein. Expression of a truncated derivative of mEH on the cell surface also establishes that mEH is integrated into the ER with more than one transmembrane domain. In addition to the bile acid transport function of mEH, the presence of various cytochrome P450s and the cytochrome P450 reductase on the hepatocyte cell surface raises provocative questions concerning the role of these proteins in extracellular carcinogen metabolism.