Interaction with PDZK1 Is Required for Expression of Organic Anion Transporting Protein 1A1 on the Hepatocyte Surface*

Although many organic anion transport protein (Oatp) family members have PDZ consensus binding sites at their C termini, the functional significance is unknown. In the present study, we utilized rat Oatp1a1 (NM_017111) as a prototypical member of this family to examine the mechanism governing its subcellular trafficking. A peptide corresponding to the C-terminal 16 amino acids of rat Oatp1a1 was used to affinity-isolate interacting proteins from rat liver cytosol. Protein mass fingerprinting identified PDZK1 as the major interacting protein. This was confirmed by immunoprecipitation of an Oatp1a1-PDZK1 complex from cotransfected 293T cells as well as from native rat liver membrane extracts. Oatp1a1 bound predominantly to the first and third PDZ binding domains of PDZK1, whereas the high density lipoprotein receptor, scavenger receptor B type I binds to the first domain. Although it is possible that PDZK1 forms a complex with these two integral membrane proteins, this did not occur, suggesting that as yet undescribed factors lead to selectivity in the interaction of these protein ligands with PDZK1. Oatp1a1 protein expression was near normal in PDZK1 knock-out mouse liver. However, it was located predominantly in intracellular structures, in contrast to its normal basolateral plasma membrane distribution. Plasma disappearance of the Oatp1a1 ligand [35S]sulfobromophthalein was correspondingly delayed in knock-out mice. These studies show a critical role for oligomerization of Oatp1a1 with PDZK1 for its proper subcellular localization and function. Because its ability to transport substances into the cell requires surface expression, this must be considered in any assessment of physiologic function.

Although many organic anion transport protein (Oatp) family members have PDZ consensus binding sites at their C termini, the functional significance is unknown. In the present study, we utilized rat Oatp1a1 (NM_017111) as a prototypical member of this family to examine the mechanism governing its subcellular trafficking. A peptide corresponding to the C-terminal 16 amino acids of rat Oatp1a1 was used to affinity-isolate interacting proteins from rat liver cytosol. Protein mass fingerprinting identified PDZK1 as the major interacting protein. This was confirmed by immunoprecipitation of an Oatp1a1-PDZK1 complex from cotransfected 293T cells as well as from native rat liver membrane extracts. Oatp1a1 bound predominantly to the first and third PDZ binding domains of PDZK1, whereas the high density lipoprotein receptor, scavenger receptor B type I binds to the first domain. Although it is possible that PDZK1 forms a complex with these two integral membrane proteins, this did not occur, suggesting that as yet undescribed factors lead to selectivity in the interaction of these protein ligands with PDZK1. Oatp1a1 protein expression was near normal in PDZK1 knock-out mouse liver. However, it was located predominantly in intracellular structures, in contrast to its normal basolateral plasma membrane distribution. Plasma disappearance of the Oatp1a1 ligand [ 35 S]sulfobromophthalein was correspondingly delayed in knock-out mice. These studies show a critical role for oligomerization of Oatp1a1 with PDZK1 for its proper subcellular localization and function. Because its ability to transport substances into the cell requires surface expression, this must be considered in any assessment of physiologic function.
A major function of the hepatocyte is the removal of various xenobiotic and endogenous organic anionic compounds from the circulation. Sulfobromophthalein (BSP) 1 is a model organic anion that circulates bound avidly to albumin and is extracted rapidly and efficiently by the hepatocyte (1)(2)(3). Establishment of a method to synthesize [ 35 S]BSP of high specific activity (4) facilitated studies to identify its hepatocyte transporter(s). Studies performed in a Xenopus laevis oocyte expression system (5) identified a candidate transporter that was initially termed organic anion transporting polypeptide (Oatp). Since this initial description more than 20 additional members of the Oatp family have been described (6,7). The original protein was termed Oatp1 and, subsequently, Oatp1a1 (see Table I) in a proposal for standardization of nomenclature (7). Studies utilizing antisense knock-out of Oatp1a1 expression in Xenopus oocytes that had been injected with rat liver mRNA suggested that this protein is responsible for a substantial fraction of organic anion transport by the liver (8), although this remains to be validated by other methods.
The family of organic anion transport proteins (Oatps) is characterized by a high degree of amino acid similarity as well as overlap of transported substrates, although their tissue distributions are varied (6,7). In addition, they have similar predicted membrane topologies and biochemical characteristics. Although evidence suggests that the Oatps are important in clearance of drugs from the circulation (6,7,9), little is known regarding the mechanism by which they act, their oligomerization state, or mechanisms for subcellular trafficking. Of note is the fact that all of the Oatps that have been examined have distinct plasma membrane distributions, except for the prostaglandin transporters in which intracellular localization appears to predominate (10). Examination of their Cterminal sequences reveals that many of the known members of the Oatp family have PDZ consensus binding sites (see Table  I). The prostaglandin transporters are among the group of Oatps that lack a putative PDZ binding domain (11,12). Generally PDZ consensus binding sites are established by the sequence of the C-terminal four amino acids (13)(14)(15). Three classes of PDZ consensus binding sites have been described, relating these peptide sequences to the crystal structures of known PDZ domains to which they bind (14). The PDZ consensus sites that are present in the hepatic Oatps are all of Class I, defined by the sequence X(S/T)X⌽, where X is any amino acid, and ⌽ is a hydrophobic amino acid (14). A relatively large number of PDZ proteins have been described (14,15), although there is as yet no way to predict which if any will bind a particular protein with a PDZ consensus binding site.
In the present study, we utilized rat Oatp1a1 (NM_017111) as a prototypical member of the Oatp family to examine * This work was supported by National Institutes of Health Grants DK23026, DK41296, and CA06576, American Heart Association Grant 0130305N, and Pfizer International High Density Lipoprotein Research Award CU516105. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18  whether interaction with a PDZ domain-containing protein provides a mechanism governing its subcellular localization. Oatp1a1 is located on the basolateral plasma membrane of the hepatocyte (16) as well as on the apical plasma membranes of the epithelial cells of the choroid plexus and the S 3 segment of the renal proximal tubule (16 -18). Its terminal four amino acids (KTKL) are consistent with a type I PDZ binding motif (14). There is also a mouse homolog of this protein (NM_013797) that is 81% identical to the rat Oatp1a1, and the C-terminal 11 amino acids of these two proteins are identical.

MATERIALS AND METHODS
Antibodies and Reagents-The antibody against the N terminus of Oatp1a1 (MEETEKKIATQEGRC) linked to KLH was prepared in rabbits by Covance Research Products Inc. (Denver, PA) as previously described (19). This antibody recognizes Oatp1a1 in rat and mouse liver and was used for immunoblots at a 1:1000 dilution. An antibody specific to rat Oatp1a1, used in immunoblots at a dilution of 1:2500, was raised in rabbits to a KLH-linked peptide corresponding to 13 amino acids near the C terminus of rat Oatp1a1 (aa 646 -658) as described previously (16). A rabbit antibody to Oatp1a4, used in immunoblots at a dilution of 1:10,000, was prepared to a KLH-linked peptide corresponding to the 11 C-terminal amino acids (aa 650 -661) of the protein.
Rabbit antibodies to scavenger receptor B type I (SR-BI) and PDZK1 were as previously described (20) and were used in immunoblots at dilutions of 1:1000. A rabbit polyclonal antibody that recognizes the mouse asialoglycoprotein receptor was kindly provided by Dr. Richard Stockert for immunofluorescence studies. ECL reagent for Western blot analysis was obtained from PerkinElmer Life Sciences. Horseradish peroxidase (HRP)-conjugated affinity-purified goat anti-rabbit IgG and HRP-conjugated affinity-purified goat anti-mouse IgG were obtained from Jackson ImmunoResearch (West Grove, PA) and were used in immunoblots at dilutions of 1:50,000 and 1:10,000, respectively. 293T cells were obtained from Dr. Robert Burk (21). All other reagents were obtained from Sigma unless otherwise noted. All animal procedures were approved by the university committees on animal use.
Preparation of C-terminal Peptide Affinity Gel-The peptides CHG-SPQVENDGELKTKL corresponding to the C-terminal 16 amino acids of rat Oatp1a1 and CHGSPQVENDGEL in which the last 4 amino acids were deleted were synthesized in The Laboratory for Macromolecular Analysis and Proteomics, Albert Einstein College of Medicine. An additional cysteine residue was included at the N terminus to facilitate coupling to Ultralink Iodoacetyl Gel (Pierce) according to the manufacturer's instructions.
Isolation of Peptide-binding Proteins from Rat Liver Cytosol-The liver was surgically removed from a rat under pentobarbital anesthesia and was immediately infused through the portal vein with 30 ml of ice-cold PBS. It was quickly weighed and Dounce-homogenized in PBS (10 ml/3 g of liver) containing protease inhibitors (Sigma, catalog # P-8340). After filtration through 3 layers of cheesecloth, the homogenate was centrifuged at 4°C at 100,000 ϫ g for 1 h. Supernatant (5 ml) was mixed with 1 ml of the peptide-coupled gel in a column and rotated overnight at 4°C. The gel was then washed successively with 200 ml of ice-cold PBS, 10 ml of 0.5 M NaCl, 10 ml of 1 M NaCl, and 20 ml of PBS. Proteins remaining bound to the washed gel were eluted with sample buffer and subjected to 10% SDS-PAGE after which they were identified by Coomassie Blue or silver-staining.
Mass Spectrometry Analysis-Coomassie Blue-stained bands were excised from SDS-polyacrylamide gels, destained with 0.2 M ammonium bicarbonate in 50% acetonitrile, and reduced using 20 mM tris (2carboxyethyl)-phosphine-HCl, and free cysteine residues were alkylated with 55 mM iodoacetamide. After digestion for 16 h at 37°C with 25 ng/l sequencing grade modified trypsin (Promega), products were cleaned and concentrated using a C 18 ZipTip (Millipore), mixed with 0.5 l of 10 mg/ml 1-cyano-4-hydroxycinnamic acid in 50% acetonitrile, 0.1% (v/v) trifluoroacetic acid, and applied onto a matrix-assisted laser desorption/ionization (MALDI) plate. Spectra were recorded with a PerSeptive Voyager-DE STR MALDI time-of-flight mass spectrometer operated in the reflection mode. The mass measurement accuracy with internal calibration was better than 100 ppm. The measured peptide masses were used for data base searching with ProFound algorithm (ProteoMetrics, NY) and Matrix Science (Mascot). For electrospray ionization MS/MS analysis, an APIQSTAR LC/MS/MS system (Applied Biosystems, Foster City, CA) was used. Fragment ion (tandem) mass spectra were obtained using collision-induced dissociation and analyzed using Matrix Science (Mascot) software.
Immunoprecipitation of Rat Liver Membrane Extracts-Antiserum against the N-terminal peptide of Oatp1a1 was immunopurified with peptide coupled to Sulfolink-agarose (Pierce) according to the manufacturer's instructions. This purified antibody was covalently coupled to immobilized protein A-agarose (Sigma) by incubating for 1 h at room temperature in 40 mM dimethyl pimelimidate (Pierce) in 0.2 M triethanolamine buffer, pH 8.2. A 0.1 M Na 2 CO 3 extracted rat liver pellet, highly enriched in Oatp1a1 (22), was resuspended in PBS containing 1% Triton X-100 and protease inhibitors (0.1 mM leupeptin, 0.1 mM 4-(2-aminoethyl)benzenesulfonyl fluoride, 0.01 mM pepstatin A, 1 mM EDTA, and 0.01% sodium azide). After centrifugation at 100,000 ϫ g for 1 h at 4°C, the supernatant was immunoprecipitated with 50 l of anti-Oatp1a1-protein A gel and washed with 1% Triton X-100 in PBS, and 200-l fractions were eluted with 0.2 M glycine, pH 2.3, directly into 1 M Tris-base to neutralize the pH. The eluates were then subjected to Western blot analysis.
Co-immunoprecipitation of Oatp1a1 and PDZK1 after Coexpression in 293T Cells-Oatp1a1 cDNA was excised from pSPORT-Oatp1a1 using the KpnI and NotI multicloning restriction enzyme sites (23) and cloned into pCDNA3.1(ϩ) (Invitrogen). PDZK1 cDNA was cloned into pFLAG-CMV-5c (Sigma) after PCR amplification from a pCDNA3.1/hygro-PDZK1 plasmid (20), resulting in a plasmid encoding PDZK1 with FLAG at its N terminus. Transient co-transfection of 293T cells with pCDNA3.1-Oatp1a1 and pFLAG-CMV/PDZK1 was performed using PolyFect transfection reagent (Qiagen Inc., Valencia, CA) according to the manufacturer's instructions. Cells were harvested 2 days after transfection, washed with PBS, and incubated for 30 -60 min on ice with PBS containing 1% CHAPS and protease inhibitors (Sigma #P-8340). The lysate was centrifuged at 100,000 ϫ g for 30 min at 4°C, and the supernatant was incubated overnight at 4°C with Oatp1a1protein A affinity gel or anti-FLAG M2 affinity gel (Sigma) (25 l of gel/4 ϫ 10 6 cells). Each gel was washed with 1% CHAPS in PBS, incubated with SDS-PAGE sample buffer, and centrifuged, and the supernatant was subjected to Western blot analysis. A control study was also performed in which Oatp1a4 and FLAG-PDZK1 were cotransfected into 293T cells, which were then subjected to FLAG immunoprecipitation as described above. For these studies, Oatp1a4 cDNA was excised from a pCR2.1-Oatp1a4 plasmid that was provided by Dr. Richard Kim (24). The Oatp1a4 insert was excised using KpnI-and XhoI-multicloning restriction enzyme sites and cloned into the pCDNA3.1/Zeo(Ϫ) plasmid.
Generation of PDZK1 Gene-targeted Mice-Mice in which exon 1 and part of intron 1 of the PDZK1 allele were replaced by the Neo cassette were prepared and bred in the Columbia University Transgenic Facility as described previously (27).
Immunofluorescence Localization of PDZK1 in Liver-Male wild type or PDZK1 knock-out mice were anesthetized with ether, and the livers were removed and fixed by immersion for 3 h at 4°C with 4% paraformaldehyde in 0.1 M phosphate buffer, pH 7.4, containing 7.5% sucrose and frozen sections (ϳ30 m thick) were prepared (16). Sections were exposed overnight at 4°C to anti-Oatp1a1, anti-asialoglycoprotein receptor, or anti-PDZK1 diluted 1:100 in PBS and, after rinsing in PBS, were exposed overnight at 4°C to a 1:400 dilution of a Cy3-labeled donkey antibody to rabbit IgG (Jackson ImmunoResearch). Controls included examination of sections for autofluorescence after exposure to nonspecific primary antiserum. Wide field immunofluorescence images were captured with a 60ϫ Olympus objective (1.4 NA) on an Olympus IX71 microscope. Rhodamine fluorescence was excited with a DG-4 (Sutter Instruments) xenon light source, and digital images were recorded on a Photometrics CoolSnap HQ CCD camera controlled by Metamorph imaging software (Universal Imaging Corp). For deconvolution, liver was optically sectioned every 0.25 m using an MS-2000 automated piezoelectric x, y, z stage (Applied Scientific Instruments) for a total of 72 optical sections. Image stacks were deconvolved using the Metamorph "measured deconvolution" algorithm. For this, suboptical resolution fluorescent beads (PS-Spec, Molecular Probes) were sectioned in the z dimension under identical conditions as the liver. Bead stacks were assembled to generate a measured "point spread function," and this point spread function was used to optically deconvolve the stacks of liver fluorescence images.

Isolation and Identification of Rat Liver Cytosolic Proteins
That Bind to the C-terminal Tail of Oatp1a1-For these experiments, a peptide corresponding to the C-terminal 16 amino acids of rat Oatp1a1 was covalently coupled to agarose gel. This peptide minus the last 4 amino acids that comprise the PDZ binding consensus domain was also coupled to agarose gel and used as a control. The columns were washed extensively with PBS as well as 1 M NaCl. Proteins bound to the washed gel were eluted with SDS-PAGE sample buffer and detected after SDS-PAGE by silver stain or staining with Coomassie Blue. A representative silver-stained SDS-PAGE gel is seen in Fig. 1a. A major protein band of ϳ70 kDa was detected only in the material that was bound to the intact peptide (lane 1) and not to the peptide lacking the terminal 4 amino acids (lane 2). This band was visualized after Coomassie Blue staining of replicate gels and was excised. After reduction and alkylation, the gel slice was incubated overnight with trypsin. The resulting tryptic peptides were identified by MALDI mass spectrometry. A representative MALDI mass spectrum is shown in Fig. 1b. Data base analysis revealed a high correspondence of the observed peptide masses to those that would be obtained by tryptic digestion of rat PDZK1. These 64 possible tryptic peptides are indicated in Fig. 1c by alternating shading that has been applied to the PDZK1 sequence. After analysis of multiple MALDI spectra, 50 peptides corresponding to 66% of the protein sequence of PDZK1 were identified as indicated by the solid underlines in Fig. 1c. Identification as PDZK1 was confirmed by tandem MS/MS analysis. A representative MS/MS spectrum is shown in Fig. 1d, in which the sequence corresponding to the fifth PDZK1 tryptic peptide was identified. Masses in this figure are annotated using standard nomenclature as described (28). Utilizing tandem MS/MS, the sequences of 27 peptides corresponding to 54% of the PDZK1 sequence were identified and are indicated by the broken underlines in Fig. 1c. Together, the two methods identified 93% of the PDZK1 sequence (Fig. 1c). In addition, identification of this protein as PDZK1 was confirmed by immunoblot using a peptide-specific antibody (data not shown).
Interaction of Oatp1a1 and PDZK1 in Cells and Rat Liver-These studies showed that the C-terminal four amino acids of Oatp1a1 are necessary for interaction with PDZK1. However, these results are not predictive as to whether this interaction between Oatp1a1 and PDZK1 actually occurs in vivo. Although several proteins have been shown to bind to PDZK1 under in vitro conditions, the functional significance of this interaction has not always been clear (29,30). To assess whether the interaction found with the C-terminal peptide occurs with fulllength proteins expressed in cells, 293T cells were transfected with expression plasmids encoding Oatp1a1 and FLAG-PDZK1. Immunoprecipitation was performed with FLAG antibody or an antibody raised to the N terminus of Oatp1a1 that would not be expected to interfere with interaction of the C terminus with PDZK1. Western blot analysis of the FLAG immunoprecipitate with anti-Oatp1a1 revealed the presence of Oatp1a1 (Fig. 2a, left panel), and Western blot analysis of the Oatp1a1 immunoprecipitate revealed the presence of FLAG-PDZK1 (Fig. 2a, right panel). There was no product detected in the immunoprecipitate after cotransfection of either expression plasmid with the alternate empty plasmid (Fig. 2a, lanes 2 and  3). As a control, coexpression of Oatp1a4 and FLAG-PDZK1 in 293T cells revealed that the FLAG-PDZK1 immunoprecipitate did not contain detectable Oatp1a4 (Fig. 2b). Rat Oatp1a4, formerly known as Oatp2, is a member of the Oatp family with an amino acid sequence that is ϳ76% identical to that of Oatp1a1. As seen in Table I, although it is distributed on the basolateral plasma membrane of the hepatocyte, it lacks a PDZ consensus sequence at its C terminus. These studies of Oatp1a1-FLAG-PDZK1 interaction were performed in cells in which synthesis of these proteins was maximized. They do not prove that such an interaction actually occurs in hepatocytes, where protein concentrations may be lower and other proteins may be competing for binding to PDZK1. Thus, although these experiments indicate that full-length PDZK1 can interact with full-length Oatp1a1, it is important to determine whether this interaction actually occurs in the liver. Consequently, a series of experiments was performed in which a rat liver membrane detergent extract was subjected to immunoprecipitation with antibody to the N terminus of Oatp1a1. The immunoprecipitates were subjected to SDS-PAGE after which Western blot analysis was performed. As seen in the left two panels of Fig.  2c, antibody to Oatp1a1 immunoprecipitated Oatp1a1 as well as PDZK1, confirming that they are bound to each other in liver. Perhaps the best characterized ligand partner for PDZK1 is the high density lipoprotein receptor, SR-BI. This protein is present on the basolateral surface of hepatocytes, where it selectively extracts lipids from high density lipoprotein particles (20,31). Previous studies have shown that SR-BI and PDZK1 coimmunoprecipitate from rat liver (25). Interestingly, as seen in the fourth panel of Fig. 2c, there was no SR-BI found in the Oatp1a1 immunoprecipitate from rat liver. Rat Oatp1a4, as noted above, lacks a PDZ consensus sequence at its C terminus. We hypothesized that it might sort to the cell surface as a complex with Oatp1a1, but as seen in the third panel of Fig.  2c, it is not present in the Oatp1a1 immunoprecipitate from rat liver. These results indicate that there is specific interaction of Oatp1a1 and PDZK1 in the liver.
PDZK1 has four independent PDZ domains (25,26). SR-BI binds to the first PDZ domain of PDZK1 (25). Experiments were designed to determine the domain(s) to which Oatp1a1 binds. Plasmids encoding PDZK1 binding domains 1 (aa 1-110), 2 (aa 113-235), 3 (aa 221-343), or 4 (aa 356 -519) as glutathione S-transferase fusion proteins were constructed. The respective proteins were expressed in E. coli and bound to GSH-agarose gels. Each gel was incubated with a Triton X-100 rat liver membrane extract. Gels were then washed exten-sively, and bound proteins were eluted into SDS-PAGE sample buffer and subjected to Western blot analysis using antibody to Oatp1a1. As seen in Fig. 2d, Oatp1a1 bound predominantly to the first and third domains of PDZK1.
Oatp1a1 Expression and Function in PDZK1 Knock-out Mice-Although the preceding studies suggest a strong interaction of Oatp1a1 with PDZK1 in the liver, they do not prove that this interaction is physiologically relevant. To examine this issue, studies were performed in mice in which expression of PDZK1 was genetically disrupted. The mouse homolog of rat Oatp1a1 has the identical PDZ recognition motif at the C terminus and is recognized by the antibody to the N terminus of Oatp1a1 that was used in the immunoprecipitation studies. Previous studies in transgenic mice revealed that interaction with PDZK1 was essential for targeting of SR-BI to the hepa-FIG. 1. Identification of rat liver cytosolic proteins that bind to the C-terminal tail of Oatp1a1. a, the peptides CHGSPQVENDGELK-TKL corresponding to the C-terminal 16 amino acids of rat Oatp1a1 (lane 1) and CHGSPQVENDGEL, in which the last 4 amino acids were deleted (lane 2), were coupled to iodoacetyl-agarose gel, incubated with rat liver cytosol, and then washed extensively with PBS and 1 M NaCl before proteins were eluted into sample buffer and resolved on 10% SDS-PAGE. A representative silver-stained gel is shown. The arrowhead indicates the major band that was further analyzed by mass spectrometry. b, a Coomassie Blue-stained band, corresponding to that indicated by the arrowhead in panel a, was excised from a replicate SDS-polyacrylamide gel and subjected to in-gel digest with trypsin. A representative MALDI-time-of-flight MS mass fingerprint is shown. Peptides corresponding to those theoretically obtained from tryptic digestion of rat PDZK1 (panel c) are indicated. c, the 64 peptides (T1-T64) that can theoretically be obtained after tryptic digestion of PDZK1 are indicated by alternating shading that has been applied to the PDZK1 sequence. Data base analysis identified multiple peptides that corresponded to the theoretical tryptic peptides of rat PDZK1. The solid underlines show peptides identified by MALDI-time of flight MS (66% of the protein sequence), and the broken underlines show peptides identified by electrospray ionization-mass spectrometry-quadrupole time-of-flight MS (54% of the protein sequence). Together, the two methods identified 93% of the rat PDZK1 sequence. d, a representative CID MS/MS product ion spectrum from quadrupole timeof-flight MS is shown. This identifies [M ϩ 2H] 2ϩ fragmentation of the T5 tryptic peptide of rat PDZK1. tocyte plasma membrane (20). In concordance with these observations, SR-BI expression and function in livers from PDZK1 knock-out mice were found to be markedly reduced (27,32). An as yet unexplained post-transcriptional process is responsible for this reduction in protein expression, as levels of mRNA encoding SR-BI were normal in these mice (32). We confirmed by Western blot that the strain of PDZK1 mice that were used in the present study had no expression of PDZK1 protein as compared with wild type mice (Fig. 3a, top panel). Similar to the earlier studies noted above that were performed in another strain of PDZK1 knock-out mice (32), this was accompanied by a substantial reduction in expression of SR-BI (Fig. 3a, middle panel). In contrast, total expression of Oatp1a1 in liver homogenate was similar to wild type when and FLAG-PDZK1 were co-transfected into 293T cells. A lysate was prepared in 1% CHAPS, and immunoprecipitations were performed using anti-FLAG coupled to agarose beads. A Western blot of 50 l of 1 ml of total lysate (Lys) and 50 l of 75 l of immunoprecipitate eluate (IP) for Oatp1a4 (arrowhead, left panel) revealed that it was not immunoprecipitated with FLAG-PDZK1, which was present in the immunoprecipitate as shown in the Western blot with FLAG antibody in the right panel. c, Na 2 CO 3 -extracted rat liver membranes were prepared and solubilized in 1% Triton X-100. After immunoprecipitation with anti-Oatp1a1 covalently coupled to agarose, bound proteins were released by 0.2 M glycine, pH 2.3, and collected directly into 1 M Tris base to neutralize the pH. Immunoblots of 50 l of 1 ml of total lysate (Lys) and 50 l of 75 l of immunoprecipitate eluate were then performed using antibodies to Oatp1a1 (C-terminal), PDZK1, Oatp1a4, or SR-BI as indicated. d, the 4 PDZ domains of PDZK1 were expressed as glutathione S-transferase fusion proteins and bound to GSH-agarose. Each gel was incubated with a 1% Triton X-100 rat liver membrane extract, and after extensive washing, bound proteins were eluted into SDS-PAGE sample buffer and analyzed by Western blot with Oatp1a1 antibody. PDZK1 domains: 1, aa 1-110; 2, aa 113-235; 3, aa 221-343; 4, aa 356 -519. Relative quantitation of the Oatp1a1 bands by densitometry is shown at the bottom of this representative gel. determined in PDZK1 knock-out mice (Fig. 3a, bottom panel). However, when subcellular distribution of Oatp1a1 was examined in liver from knock-out mice, it was found to be predominantly in intracellular structures, in contrast to the typical basolateral distribution of Oatp1a1 (16) in wild type mouse liver (Fig. 3b, top panels). Importantly, the absence of PDZK1 did not cause a general disturbance in membrane behavior since the basolateral distribution of the asialoglycoprotein receptor in livers from normal and knock-out mice were indistinguishable (Fig.  3b, middle panels). As expected, PDZK1 was undetectable in the knock-out and had a primarily basolateral distribution in the wild type liver (Fig. 3b, bottom panels). Much of the cytosolic pool of PDZK1 likely washed out during the preparation and permeabilization of the tissue.
The functional significance of this subcellular redistribution of hepatocyte Oatp1a1 was determined by quantifying the plasma disappearance of [ 35 S]BSP. BSP is a well characterized substrate for Oatp1a1-mediated cellular uptake (8,9,22), although several other hepatocyte plasma membrane proteins also have the ability to mediate its uptake (9). This probably accounts for the fact that plasma disappearance of BSP was relatively rapid in both wild type and PDZK1 knock-out mice (Table II). Analysis of these data by non-linear least squares regression revealed that the plasma volume of distribution was identical in wild type and knock-out mice (Table II). However, the fractional uptake rate of BSP was reduced by ϳ25% (p Ͻ 0.05), and the corresponding plasma half-life was increased by the same proportion (p Ͻ 0.05) in the PDZK1 knock-out as compared with wild type mice (Table II).   ). b, immunofluorescence for Oatp1a1, the asialoglycoprotein receptor (ASGPR), and PDZK1 was performed on sections of wild type or PDZK1 (Ϫ/Ϫ) livers and analyzed after computer deconvolution. Single representative images from the deconvolved stacks are presented. In the liver from PDZK1 (Ϫ/Ϫ) mice, Oatp1a1 is expressed only intracellularly in contrast to the basolateral distribution in liver from PDZK1 (ϩ/ϩ) mice. Scale bar ϭ 20 m.

DISCUSSION
The present study establishes that Oatp1a1 binds to PDZK1 in vitro and in vivo. PDZK1 is a 70-kDa protein with 4 independent PDZ domains that has been shown to be present in a number of tissues including liver and kidney (20,25,26,32,33). Although several proteins have been shown to bind to PDZK1 under in vitro conditions, the functional significance of this interaction has not always been clear (30). Perhaps the best characterized ligand partner for PDZK1 is the high density lipoprotein receptor, SRBI. SRBI and PDZK1 coimmunoprecipitate from rat liver, and studies in transgenic mice reveal that this interaction is essential for targeting of SRBI to the hepatocyte plasma membrane (20). In concordance with these observations, there is markedly reduced SRBI expression and function in livers from PDZK1 knock-out mice (27,32). The present study indicates that interaction with PDZK1 is also essential for hepatocyte plasma membrane expression of Oatp1a1. SRBI binds to the first PDZ domain of PDZK1 (25), whereas Oatp1a1 binds predominantly to the first and third domains (Fig. 2d). Although it is, thus, possible that PDZK1 could form a complex with these two membrane proteins, we have found no evidence for this (Fig. 2c), suggesting that as yet undescribed factors lead to selectivity in the interaction of these protein ligands with PDZK1.
Although mouse Oatp1a4 has been described as the homolog of rat Oatp1a4 (34), the mouse protein has a C-terminal PDZ binding consensus site (KTKL), whereas the rat protein does not (Table I). These proteins are 89% identical, although the C terminus of the mouse protein is eight amino acids longer than that of the rat protein. Despite the lack of a PDZ binding domain, the rat protein localizes to the basolateral plasma membrane of the hepatocyte (35). Although it is possible that Oatp1a4 traffics to the plasma membrane as a complex with Oatp1a1, we found no evidence for their binding to each other as determined by failure to recover Oatp1a4 after immunoprecipitation of Oatp1a1 from rat liver lysate (Fig. 2c). As shown in the present study, expression of Oatp1a1 at the plasma membrane requires its interaction with PDZK1 and apparently cannot utilize the as yet undescribed PDZ-independent mechanism that is utilized by Oatp1a4.
As noted above, the prostaglandin transporters, members of the Oatp family, also lack PDZ consensus binding sites. Immunolocalization of prostaglandin transporter in various rat tissues reveals a predominant intracellular distribution (10), whereas Oatp1a1 is predominantly on the plasma membrane of hepatocytes, renal tubules, and choroid plexus epithelial cells (16 -18). It is interesting that in the choroid plexus, Oatp1a1 is distributed in large intracellular vesicular structures during the initial 8 -10 weeks of development before assuming the adult apical plasma membrane phenotype (18). Whether this is a result of altered expression of PDZK1 during development is not known at the present time.
The present study shows that interaction of Oatp1a1 with PDZK1 is required for its expression on the hepatocyte surface. As its ability to transport substances into the cell requires surface expression, this must be considered in any assessment of its physiologic function. Most studies of Oatp1a1 function have been performed in transfected cells lacking PDZK1, in which the transporter is overexpressed. Whether the lack of a PDZK1 binding scaffold will have an affect on Oatp1a1-mediated transport function is an important subject that will need to be addressed in future studies.