ARTICLES J Am Soc Nephrol 10: 685–695, 1999

Evidence for an Essential Role of Megalin in Transepithelial Transport of

ERIK ILSØ CHRISTENSEN,* JAN ØIVIND MOSKAUG,‡ HENRIK VORUM,† CHRISTIAN JACOBSEN,† THOMAS E. GUNDERSEN,‡ ANDERS NYKJÆR,§ RUNE BLOMHOFF,‡ THOMAS E. WILLNOW§ and SØREN K. MOESTRUP† *Department of Cell Biology, Institute of Anatomy and †Department of Medical Biochemistry, University of Aarhus, Denmark; ‡Institute for Nutrition Research, University of Oslo, Norway; and §Max-Delbrueck-Center for Molecular Medicine, Berlin, Germany.

Abstract. Transepithelial transport of retinol is linked to reti- urinary excretion of RBP and retinol, demonstrating that glo- nol-binding (RBP), which is taken up and also synthe- merular filtered RBP-retinol of megalin-deficient mice escapes sized in a number of epithelia. By immunocytochemistry of uptake by proximal tubules. A direct megalin-mediated uptake human, rat, and mouse renal proximal tubules, a strong staining of purified RBP-retinol was indicated by surface plasmon in apical endocytic vacuoles, lysosomes, endoplasmic reticu- resonance analysis and uptake in immortalized rat yolk sac lum, Golgi, and basal vesicles was observed, in accordance cells. Uptake was partially inhibited by a polyclonal megalin with luminal endocytic uptake as well as a constitutive syn- and the -associated protein. The present data thesis and basal secretion of RBP. Analysis of mice with target show that the absence of RBP-binding megalin causes a sig- disruption of the gene for the major endocytic receptor of nificantly increased loss of RBP and retinol in the urine, proximal tubules, megalin, revealed no RBP in proximal tu- demonstrating a crucial role of megalin in vitamin A homeosta- bules of these mice. Western blotting and HPLC of the urine of sis. the megalin-deficient mice instead revealed a highly increased

Retinol-binding protein (RBP1) is a 21-kD plasma protein and retinol subsequent to tubular uptake is being returned to the the main carrier of vitamin A (retinol) in plasma. Retinol is circulation again in complex with RBP. coupled to RBP in the and the complex is circulating in Because RBP is taken up in proximal tubules, and retinol plasma bound to (TTR), previously named preal- stimulates the expression of megalin in a rat proximal bumin, which to a certain extent prevents the RBP-retinol tubule cell line (10), we investigated if this highly expressed complex from being filtered in the glomeruli. However, 4 to receptor in proximal tubules might mediate uptake of RBP. 5% of the circulating RBP-retinol complex is not bound to Megalin, a 600-kD protein localized in the endocytic pathway TTR (1), and the kidney appears to be a very important organ of renal proximal tubules (11), belongs to the LDL receptor in the recycling of RBP-retinol, contributing about 50% of the family (12–14). The protein functions as an endocytic receptor total circulating pool of RBP-retinol as estimated in rat (2). In for a wide variety of substances, including lipoproteins (15– accordance with this, RBP has been intensively used clinically 18), (19), and basic drugs (20) (reviewed in reference to determine proteinurias of tubular origin (3,4), further indi- (21)). In addition, megalin binds calcium and receptor-associ- cating that RBP is filtered to a major extent in the renal ated protein (RAP), a chaperone-like protein (22). Interest- glomeruli and taken up in proximal tubules as confirmed by ingly, megalin also functions as a receptor for endocytic uptake immunohistochemistry (5,6). The role of the kidney in retinol of the two vitamin carrier , vitamin B12-carrier transco- homeostasis is also substantiated by the finding that acute renal balamin (23) and vitamin D-binding protein (A. Nykjær, D. failure induces significant elevations of serum retinol concen- Dragun, D. Walther, et al., Cell 1999, in press). These findings trations (7,8). Furthermore, RBP mRNA has been detected in demonstrate that megalin is fundamental for the retainment of rat kidney by in situ hybridization (9), and it is suggested that substances vital for the organism. The general importance of megalin is supported by the findings that knockout of the megalin gene in mice results in a high mortality and develop- Received September 23, 1998. Accepted October 5, 1998. mental abnormalities (24). Correspondence to Dr. Erik Ilsø Christensen, Department of Cell Biology, The present study was carried out to investigate the impor- Institute of Anatomy, University of Aarhus, DK-8000 Aarhus C, Denmark. Phone: (45) 89 42 30 57; Fax: (45) 86 19 86 64; E-mail: [email protected] tance of megalin for renal proximal tubular reabsorption and synthesis of RBP, which is suggested to control the transepi- 1046-6673/1004-0685$03.00/0 Journal of the American Society of Nephrology thelial transport of retinol. Using different approaches includ- Copyright © 1999 by the American Society of Nephrology ing immunocytochemistry and urine analyses of megalin-defi- 686 Journal of the American Society of Nephrology J Am Soc Nephrol 10: 685–695, 1999

Figure 1. Immunohistochemical labeling of retinol-binding protein (RBP) in rat (A and B) and in human (C and D) renal cortex. (A) An intense labeling is seen in a cross-sectioned segment 1 (S1), including a granular and an apparent cytoplasmic labeling. One cell (arrow) shows a pronounced cytoplasmic labeling. In addition, these cells demonstrate labeling of several basal granules (arrowheads). A segment 2 proximal tubule (S2) shows only granular, probably lysosomal, labeling. (B) Granular labeling of segment 2 (S2) proximal tubules. One cell shows in addition intense cytoplasmic labeling (arrow). Distal tubule (DT) is unlabeled. (C) Section from human cortex demonstrates proximal tubules with intense granular labeling including basal granules (arrowheads) and varying cytoplasmic labeling. (D) Several interstitial fibroblast-like cells (arrows) exhibit an intense cytoplasmic labeling. Glomerulus (G) is unlabeled. Magnification: ϫ1000 in A and C; ϫ750 in B and D.

cient mice and normal control animals, uptake studies in buffered with NH3, pH 10, as described (25). Purified RBP was then megalin-expressing cells, and ligand binding to purified mega- analyzed by sodium dodecyl sulfate-polyacrylamide gel electrophore- lin, we now evidence a key function of megalin in this process. sis, together with known amounts of commercially available RBP (Sigma, St. Louis, MO). The gel was stained with Coomassie R-250, and the concentration of purified RBP was estimated visually by the Materials and Methods Ϫ Ligands and intensity of the bands. RBP was stored frozen at 20°C. A rabbit polyclonal antihuman RBP antibody and a rabbit poly- RBP was purified from human plasma from Blodbanken, Ullevål clonal antihuman cathepsin D antibody were obtained from Dako sykehus, Oslo, Norway. Plasma was first dialyzed against 0.05 M (Dako A/S, Glostrup, Denmark), mouse monoclonal antihuman RBP NaH2PO4 containing 0.05 M NaCl, pH 7.5, to separate RBP-retinol was from Transduction Laboratories (Lexington, KY), and rat mono- from TTR at low ionic strength. The dialyzed plasma was then loaded clonal anti-mouse LAMP2 against lysosome-associated membrane onto a diethylaminoethyl-Sepharose Fast Flow column (Pharmacia, protein was a kind gift from Dr. Ira Mellman (Department of Cell Uppsala, Sweden) and eluted by a NaCl gradient, 0.05 to 0.6 M using Biology, Yale University School of Medicine, New Haven, CT). an fast protein liquid chromatography (Pharmacia). The fractions absorbing at 313 nm containing RBP were collected, pooled, and submitted to gel filtration in phosphate-buffered saline on a Superdex Preparation of Renal Tissue 200 preparative column using the fast protein liquid chromatography. Normal, uninvolved human renal tissue was obtained from resected The fractions absorbing at 313 nm were pooled and loaded onto a renal carcinoma kidneys and fixed in 8% paraformaldehyde in 0.1 M TTR (purchased from Scigen, Kent, United Kingdom)-coupled sodium cacodylate buffer, pH 7.2. Mouse megalin knockout and Sepharose at 140 mM NaCl and eluted at low ionic strength in water control kidneys were fixed by perfusion through the with 4% J Am Soc Nephrol 10: 685–695, 1999 Megalin-Mediated Endocytosis of RBP 687

Figure 2. Immunohistochemical labeling for RBP in wild-type mice (A and B) and in megalin-deficient mouse (C). (A) Intense granular labeling (arrowhead) of initial part of proximal tubule; arrow points to transition from Bowman’s capsule to proximal tubule epithelium. (B) Granular labeling of early proximal tubule (arrowhead); surrounding cross sections of proximal tubules are unlabeled. (C) Section from megalin-deficient mouse; no labeling is observed. Arrows indicate start of proximal tubule. (D) Section from wild-type mouse incubated without primary antibody shows no labeling. Arrow indicates start of proximal tubule. Magnification, ϫ1000.

paraformaldehyde in the same buffer, and rat kidneys were fixed by Ultracut E ultramicrotome. For immunolabeling, the sections were retrograde perfusion through the abdominal aorta with 1 or 4% para- incubated with primary polyclonal anti-RBP diluted 1:800 to 1:4000 formaldehyde. The tissue was trimmed into small blocks, further fixed or monoclonal anti-RBP (1 to 10 ␮g/ml) either at room temperature by immersion for1hin1%paraformaldehyde, infiltrated with 2.3 M for1horovernight at 4°C after preincubation in phosphate-buffered sucrose containing 2% paraformaldehyde for 30 min, and frozen in saline containing 0.05 M glycine and 0.1% nonfat dry milk or 1% liquid nitrogen. For some electron microscope immunocytochemical bovine . For electron microscopic double labeling experiments, tissue treated and frozen as above was further freeze- experiments, sections from human kidneys were incubated with substituted in a Reichert AFS (Reichert, Vienna, Austria) as follows: monoclonal anti-RBP together with rabbit anti-cathepsin D, 1:5000 3 d in methanol containing 0.5% uranyl acetate at Ϫ80°C, washed in (cryosections) or 1:400 (HM20 embedded tissue). Sections from methanol for 2 d increasing the temperature gradually to Ϫ45°C, mouse kidneys were incubated with polyclonal anti-RBP together gradually infiltrated over 2 d with Lowicryl HM20 (Polysciences, with rat monoclonal anti-LAMP2 diluted 1:10 to 1:100. For light Waldkraiburg, Germany) and methanol 1:2, 1:1, and finally pure microscopy, the sections were subsequently incubated with peroxi- Lowicryl HM20, and then polymerized by ultraviolet polymerization dase-conjugated secondary antibodies (Dako), the peroxidase was at Ϫ45°C for2dandat0°Cfor2dasdescribed previously (26). visualized with diaminobenzidine, and the sections were subsequently counterstained with Meier’s stain for 2 min and examined in a Leica Immunocytochemistry DMR microscope (Wetzlar, Germany) equipped with a Sony 3CCD For light microscopy, 0.8-␮m cryosections were obtained at color video camera attached to a Sony Digital Still recorder. Images Ϫ80°C, and for electron microscopy ultrathin (70 to 90 nm) sections were processed using Adobe Photoshop 4.0 and printed on Kodak were obtained at Ϫ100°C with an FCS Reichert Ultracut S cryoultra- Ektotherm XLS paper with a Kodak ColorEase PS printer. For elec- microtome as described previously (22). Ultrathin (400 to 600 nm) tron microscopy (after the primary antibody incubation), the sections sections were obtained from the cryosubstituted tissue with a Reichert were incubated with 5 or 10 nm goat anti-rabbit, goat anti-mouse, or 688 Journal of the American Society of Nephrology J Am Soc Nephrol 10: 685–695, 1999

Figure 3. Electron microscope immunocytochemistry for RBP on rat proximal tubules from cryosubstituted tissue, using polyclonal anti-RBP and 10 nm gold. (A) Two intensely labeled lysosomes (L) are seen; mitochondria (M) and cytoplasm is almost devoid of background. (B) Small vesicles (V) in the basal cytoplasm are located close to the basolateral plasma membrane. BM, basement membrane. Magnification: ϫ35,000 in A and ϫ63,000 in B.

goat anti-rat gold particles (BioCell, Cardiff, United Kingdom). The column for separation and detected by ultraviolet absorption at 326 cryosections were embedded in 2% methylcellulose containing 0.3% nm. The urine from control mice was spiked with retinol to confirm uranyl acetate, and the sections from cryosubstituted tissue were the identity of the retinol peak. stained with uranyl acetate for 10 min and studied in a Philips CM100 electron microscope. Surface Plasmon Resonance Analysis Controls For the surface plasmon resonance analyses, the BIAcore sensor chips (type CM5, Biosensor, Uppsala, Sweden) were activated with a Sections were incubated with secondary antibodies alone or with 1:1 mixture of 0.2 M N-ethyl-NЈ-(3-dimethylaminopropyl) carbodi- nonspecific rabbit or mouse IgG. None of the controls showed any imide and 0.05 M N-hydroxysuccinimide in water according to the labeling. manufacturer’s instructions. Rabbit renal megalin was immobilized as described (27). The surface plasmon resonance signal from immobi- Analysis of Megalin Gene Knockout Mice lized rabbit megalin generated 15 to 21 ϫ 103 BIAcore response units Mice genetically deficient for megalin were generated by targeted (RU) equivalent to 25 to 35 fmol/mm2. The flow cells were regener- gene disruption as described previously (24). Most newborn knockout ated with 20 ␮l of 1.5 M glycine-HCl, pH 3.0. The flow buffer was 10 mice die from developmental defects primarily affecting the forebrain mM Hepes, 150 mM NaCl and 1.5 mM CaCl2 and 1 mM ethylene- and the lung. However, approximately 1 out of 50 receptor-deficient glycol-bis(␤-aminoethyl ether)-N,NЈ-tetra-acetic acid, pH 7.4. The animals survives to adulthood. These animals were used to study the binding data were analyzed using the BIAevaluation program. The consequence of the receptor defect for tubular function. For urine number of ligands bound per immobilized receptor was estimated by collection, mice were placed in metabolic cages for 16 h and given dividing the ratio RUligand/Massligand with RUreceptor/Massreceptor. 10% sucrose in the drinking water. Urine samples obtained were qualitatively indistinguishable from samples collected without sucrose 125 load. Urine volume per hour and creatinine levels were identical in Internalization and Degradation of I-RBP by Brown megalin Ϫ/Ϫ and in control mice (data not shown). Urinary excretion Norway/Mouse Sarcoma Virus Cells of RBP was analyzed after subjecting the urine samples to 4 to 15% Megalin-expressing Brown Norway rat yolk sac epithelial cells nonreducing sodium dodecyl sulfate-polyacrylamide gel electrophore- transformed with mouse sarcoma virus (28) were grown to confluence sis. Samples were transferred to nitrocellulose filters and incubated (approximately 400,000 cells/well) in 24-well plates (Nunc A/S, Den- with 5 ␮g/ml rabbit antihuman RBP antibody. Bound IgG was de- mark) in minimum essential medium (Life Technologies, Paisley, tected with the enhanced chemiluminescence system (ECL; Amer- United Kingdom) containing 10% fetal calf serum. RBP saturated sham). with retinol was purified from human plasma as described. Iodination Urine collected from control mice and megalin-deficient mice was was performed by the chloramine-T method. The specific activity was analyzed for retinol by HPLC as described by Gundersen et al. (25). approximately 1 MBq/␮g. Incubation with 125I-RBP was carried out Briefly, 40 ␮l of urine was extracted on-line on a solid-phase C18 in minimum essential medium supplemented with 0.1% bovine serum column using a column switching system before the analytes bound to albumin. Degradation of the proteins was measured by precipitation of the pre-column were backflushed to the Suplex pKb-100 analytical the incubation medium in 12.5% TCA. Cell-associated radioactivity J Am Soc Nephrol 10: 685–695, 1999 Megalin-Mediated Endocytosis of RBP 689

Figure 4. Double labeling electron microscope immunocytochemistry for RBP (mouse monoclonal anti-RBP and 5 nm gold) and cathepsin D (polyclonal anti-cathepsin D and 10 nm gold) on cryosections from human kidney. Two lysosomes (L) in the basal part of a proximal tubule cell labeled for cathepsin D (arrows) are also intensely labeled for RBP (arrowheads). A cytoplasmic body (CB) is labeled exclusively for RBP (arrowheads). Inset. Lysosome (L) intensely labeled for cathepsin D (arrows) and RBP (arrowheads). Magnification, ϫ86,000.

was measured by counting the cells after solubilization of the cell medulla (not shown). Mice (Figure 2, A and B) showed a layer in 1% Triton X-100. staining similar to that described in the rat, whereas no labeling for RBP was seen in five megalin-deficient mice analyzed Results (Figure 2C). Sections from control mice incubated without Immunocytochemistry primary antibody (Figure 2D) showed no labeling. Light microscope immunohistochemistry of rat and human Electron microscope immunocytochemistry of cryosections proximal tubules revealed a strong granular staining for RBP of from the three species revealed an intense labeling of endo- especially the early segments of the proximal tubule. In some somes and especially of lysosomes as shown in Figures 3 rats and in the three human kidneys studied, it also extended through 5 and verified by double labeling for cathepsin D in into segment 2 as seen in Figure 1, A through C. In two of the human (Figure 4) and LAMP2 in mice (Figure 5A). Basal human kidneys, there was also a staining in segment 3 (not shown). In addition to the granular staining corresponding to cytoplasmic granules of all species were also labeled for RBP the location of endosomes and lysosomes, there was also, in (Figures 3B), and double labeling for cathepsin D in human both rat and human kidneys, a staining of small granules (Figure 4) and LAMP2 in mice (Figure 5, B and C) indicated located close to the basement membrane (Figure 1, A and C). that these vesicles were not lysosomes. Furthermore, there Furthermore, a distinct apparent cytoplasmic labeling was ev- appeared in these cells to be labeling of the granular endoplas- ident. This labeling was seen mainly in the early parts of the mic reticulum and vesicles in the Golgi region for RBP (Figure proximal tubule (Figure 1A), but also in occasional single cells 6). Altogether, these data suggest an apical endocytic uptake as further down in the proximal tubule (Figure 1B). In the human well as synthesis and basolateral secretion of RBP. kidneys, there was a strong labeling of interstitial fibroblast- The labeling of the interstitial fibroblast-like cells was con- like cells in the cortex (Figure 1D) and outer zone of the outer firmed at the electron microscope level. These cells exhibited 690 Journal of the American Society of Nephrology J Am Soc Nephrol 10: 685–695, 1999

Figure 5. Double labeling electron microscope immunocytochemistry for RBP (polyclonal anti-RBP and 5 nm gold) and LAMP2 (rat monoclonal anti-LAMP2 and 10 nm gold) on cryosection (A) or sections from cryosubstituted tissue (inset to A) from wild-type mouse proximal tubules. Panel A and inset to A show lysosomes labeled for LAMP2 (arrows) and RBP (arrowheads). (B and C) Small cytoplasmic bodies located in the basal part of the cell and often close to the plasma membrane are labeled exclusively for RBP (arrowheads). Magnification: ϫ86,000 in A, B, and C; ϫ63,000 in inset to A. a very high cytoplasmic labeling that appeared not to be If the excreted RBP was in the ligand-bound form it would also confined to vacuolar structures (not shown). suggest that absence of megalin causes loss of vitamin A in the urine. To test this, we analyzed the urine for the presence of Urine Analysis of Megalin Gene Knockout Mice retinol by HPLC. Retinol was detected in the urine from the Analysis of the urine from four control mice and three megalin-deficient mice at an estimated concentration of 0.2 to megalin-deficient mice revealed several additional Coomassie 0.4 ␮M, whereas retinol was not detectable in the urine from blue-stained bands in knockout urine samples compared with normal mice (Figure 8). controls, and Western blot analysis demonstrated that only the three megalin-deficient mice excreted RBP in the urine Surface Plasmon Resonance Analysis (Figure 7). Surface plasmon resonance analysis was used for investiga- This observation (Figure 7) strongly suggests that megalin is tion of the binding of purified RBP to megalin (Figure 9). The important for reabsorption of RBP from the glomerular filtrate. analysis revealed that RBP binds to megalin, although the J Am Soc Nephrol 10: 685–695, 1999 Megalin-Mediated Endocytosis of RBP 691

Figure 6. Electron microscope immunocytochemistry for RBP on rat proximal tubules from cryosubstituted tissue, using polyclonal anti- RBP and 10 nm gold. (A) Labeling for RBP is seen in rough endoplasmic reticulum (arrows), in a small vesicle (V), and in a tubulovesicular structure (arrowheads). (B) Golgi apparatus (G). Labeling is observed in small Golgi vesicles (arrows). Magnification: ϫ86,000 in A and ϫ63,000 in B. interaction of the purified ligand and receptor has a relatively megalin. 125I-labeled RBP was effectively taken up, and radio- low affinity (1 ␮M at 20°C). The mass equivalent response active degradation products appeared in the medium increasing units generated by RBP were 10 times less than that of RAP. with time. The uptake was partially inhibited by RAP and a This is probably due to the difference in mass of RBP (21 kD) polyclonal antibody raised against megalin (Figure 10), compared with recombinant RAP (37 kD), and the high num- whereas no significant effect was seen with a nonimmune ber of RAP binding sites in megalin rather than difference in antibody. affinity. When RAP was bound before binding to RBP, no significant binding to RBP could be measured. Discussion Internalization and Degradation of 125I-RBP by Brown The present study reveals that megalin is a receptor for Norway/Mouse Sarcoma Virus Cells tubular reabsorption of RBP. Since 85 to 90% of the circulating Endocytosis of 125I-RBP was studied (Figure 10) in a rat RBP is saturated with retinol (29), megalin-mediated uptake of yolk sac epithelial cell line that exhibits a high expression of RBP therefore prevents urinary loss of retinol under physio- 692 Journal of the American Society of Nephrology J Am Soc Nephrol 10: 685–695, 1999

Figure 7. Analysis of RBP excretion in mouse urine. Fifteen microliters of urine from mice of the indicated megalin genotypes was subjected to 4 to 15% nonreducing sodium dodecyl sulfate-polyacrylamide gel electrophoresis and staining with Coomassie blue (lanes 1 through 7). Duplicate samples (lanes 8 to 14) were transferred to nitrocellulose filters and incubated with 5 ␮g/ml rabbit antihuman RBP antibody. Bound IgG was detected with the enhanced chemiluminescence system (ECL; Amersham). The position of migration of molecular weight marker proteins in the gel is indicated. logic conditions. Furthermore, the results suggest that RBP is substantial amounts of RBP. The MDCK II cells were found to degraded in lysosomes subsequent to the endocytic uptake. secrete RBP with bound retinol preferentially to the basolateral Retinol may then be coupled to newly synthesized RBP, which side when cells were grown on permeable support. Since the carry the vitamin back to the circulation. kidney appears to be a major source of recirculating RBP- RBP is a widely used marker for proteinuria of proximal retinol complexes and since reabsorbed RBP has probably been tubular origin (30), confirming that this protein is normally degraded in the lysosomes (Figures 3 through 5 and Figure 10), reabsorbed in this part of the nephron subsequent to glomerular it seems most likely that the released retinol is being coupled filtration. Endogenous RBP reabsorbed from the tubular fluid to newly synthesized RBP in the same cells. Actually, previous has also been identified in proximal tubules by immunohisto- chemistry in human kidneys (6) and in rat (5). We found no labeling of segment 3 in rat and mouse in agreement with the findings of Kato et al. (5). However, in two of the human kidneys studied, there was a significant labeling of segments 3 in the outer medulla. These differences in the segmental dis- tribution of RBP might be due to differences in vitamin A intake between individuals and thereby the amounts of RBP presented to the proximal tubule. The subcellular pattern of immunolabeling for RBP in the proximal tubule in all three species studied, i.e., the apparent cytoplasmic labeling, is quite different from other endogenous proteins reabsorbed in the proximal tubule, including, e.g., ␤ 2-glycoprotein-I (15) or albumin (E. I. Christensen, unpub- lished observations), which probably reflects that these pro- teins are not synthesized in the epithelium. The localization of RBP in rough endoplasmic reticulum, Golgi, and basal nonlysosomal granules often located close to the basolateral plasma membrane strongly suggests a cellular synthesis and basal secretion of RBP. Northern blot analysis Figure 8. HPLC analysis of urine from megalin knockout mouse. has demonstrated RBP mRNA in the kidney at a level of 5 to Urine samples from megalin knockout and control mice were col- lected, and 40 ␮l was injected into the HPLC system. The samples 10% of that of the liver (31), and synthesis of RBP in proximal were extracted on-line on a solid-phase column and chromatographed tubule is also suggested from studies by Makover et al. (9), on a C18 Bondapak column. The separated substances were detected although these authors found RBP mRNA preferentially in with an ultraviolet detector at 326 nm. Identification of the retinol segment 3 by in situ hybridization. Furthermore, recent studies peak was done by mixing urine from a control animal with retinol by Moskaug (unpublished data) on cultured kidney cell lines standard and subsequent chromatography with identical conditions as LLCPK1 and MDCK II have shown synthesis and secretion of with urine from megalin knockout mouse. J Am Soc Nephrol 10: 685–695, 1999 Megalin-Mediated Endocytosis of RBP 693

Figure 9. SPR sensorgram of the binding of RBP to megalin and megalin- receptor-associated protein (RAP) complex. The upper curve shows the binding of RAP (40 ␮g/ml). After dissociation of the loosely bound RAP (approximately 50% of total bound), RBP (10 ␮g/ml) was exposed to the sensor chip. Only a weak increase in response is seen compared with the response by exposure of RBP to the same megalin chip without prior binding of RAP (lower curve). The curves and values displayed represent the response after subtrac- tion of nonspecific binding, which was measured to a megalin-chip inactivated by reduction.

125 studies have suggested the same mechanism for transcellular Figure 10. Uptake and degradation of I-RBP in the rat yolk sac transport of retinol in the yolk sac of rodents (32–34), which carcinoma cell line Brown Norway/mouse sarcoma virus. (A) Time course for the uptake (E) and degradation (F). Ordinate, fraction of bear strong similarities to proximal tubular epithelium, includ- added 125I-RBP. (B) The inhibitory effect on uptake by RAP (1 ␮M), ing a high expression of megalin (35,36). sheep anti-rat megalin IgG (100 ␮g/ml), and sheep nonimmune IgG It is clear from the results shown in Figure 8 that the lack of (100 ␮g/ml) after incubation for 2 h. Confluent cell layers in 24-well megalin in the proximal tubule causes loss of substantial plates were incubated with 125I-RBP in 400 ␮l of minimum essential amounts of retinol. A similar lack of megalin in yolk sac could medium and 0.1% . Degradation was measured in part be responsible for the observed embryonic defects of the as the increase in TCA-soluble radioactivity in the medium. Uptake megalin knockout mice (24). Thus, retinoic acid deficiency in was measured as the cell-associated radioactivity plus the degraded embryos induced by inhibition of yolk sac synthesis of RBP fraction. also induces embryonic defects such as malformations of vitelline vessels, the cranial neural tube, and the eye (33). The lack of RBP staining in the proximal tubules of megalin proximal tubule cells. This accumulation of RBP has not been knockouts further suggests that retinol itself is essential for the described before, and it is tempting to speculate that these cells induction of RBP synthesis in these cells, as also suggested by represent a kidney counterpart to the hepatic stellate cells (also the observation that treatment of F9 embryonal carcinoma cells named lipid-laden cells or Ito cells) believed to represent a with retinoic acid induces differentiation into embryoid bodies hepatic storage pool for retinol and RBP (2). In this context, it and synthesis and secretion of RBP and TTR (37), although is interesting that Nagy et al. (40) recently described what was this seems not to be the case in yolk sac (38) or in liver (39). called kidney stellate cells, in kidneys from rats fed a diet with In this context, it is also noteworthy that retinol induces a high high vitamin A content. These cells accumulate retinyl esters in increase in megalin expression and megalin-mRNA in F9 cells, lipid droplets that can be visualized by endogenous fluores- as well as in a rat kidney proximal tubule cell line and in JEG-3 cence from retinyl esters (40) cells (10). Although RAP virtually abolished binding of RBP to mega- None of the species studied exhibited any labeling of other lin in the surface plasmon resonance analysis experiments, nephron segments or collecting ducts. However, the human there was only partial inhibition of RBP internalization with kidney showed a significant labeling of interstitial fibroblast- RAP and antimegalin antibodies in Brown Norway/mouse like cells in the cortex and also in the outer zone of the outer sarcoma virus cells. These cells (41) originate from yolk sac, medulla of the kidneys, showing labeling of segment 3 of the and the RBP-receptor p63, originally described in the retinal 694 Journal of the American Society of Nephrology J Am Soc Nephrol 10: 685–695, 1999 pigment epithelium (42), was recently found in mouse yolk sac 13. Saito A, Pietromonaco S, Loo AK, Farquhar MG: Complete (34). This receptor therefore may in part be responsible for the cloning and sequencing of rat gp330/“megalin,” a distinctive RBP uptake in the Brown Norway/mouse sarcoma virus cells. member of the LDL receptor gene family. Proc Natl Acad Sci We have shown previously that megalin is responsible for USA 91: 9725–9729, 1994 14. Hja¨lm G, Murray E, Crumley G, Harazim W, Lundgren S, the tubular reabsorption of -B12 (23) and for vitamin D binding protein (A. Nykjær, D. Dragun, D. Walther Onyango I, Ek B, Larsson M, Juhlin C, Hellman P, Davis H, Åkerstro¨m G, Rask L, Morse B: Cloning and sequencing of et al., Cell 1999, in press). Together with the results of the ϩ human gp330, a Ca2 -binding receptor with potential intracel- present article, these findings not only illustrate the importance lular signaling properties. Eur J Biochem 239: 132–137, 1996 of megalin as a general endocytic receptor for protein in the 15. Moestrup SK, Schousboe I, Jacobsen C, Leheste J-R, Chris- proximal tubule (21), but also emphasize a multifaceted role of tensen EI, Willnow TE: Beta2-glycoprotein-I (apolipoprotein H) megalin in retaining and capturing vital substances from the and beta2-glycoprotein-I-phospholipid complex harbor a recog- tubular fluid after glomerular filtration. nition site for the endocytic receptor megalin. J Clin Invest 102: 902–909, 1998 Acknowledgments 16. Willnow TE, Goldstein JL, Orth K, Brown MS, Herz J: LDL This work was supported by grants from the Danish Medical receptor-related protein and gp330 bind similar ligands, includ- Research Council, the Novo Nordic Foundation, the Danish Biotech- ing plasminogen activator-inhibitor complexes and , nology Program, and University of Aarhus Research Foundation. The an inhibitor of chylomicron remnant clearance. J Biol Chem 267: authors thank Hanne Sidelmann and Inger Blenker Kristoffersen for 26172–26180, 1992 excellent technical assistance. 17. Kounnas MZ, Loukinova EB, Stefansson S, Harmony JA, Brewer BH, Strickland DK, Argraves WS: Identification of glycoprotein 330 as an endocytic receptor for apolipoprotein References J/clusterin. J Biol Chem 270: 13070–13075, 1995 1. Green MH, Green JB: Dynamics and control of plasma retinol. 18. Stefansson S, Chappell DA, Argraves KM, Strickland DK, Ar- In: Vitamin A in Health and Disease, edited by Blomhoff R, New graves WS: Glycoprotein 330/LDL receptor-related protein-2 York, Marcel Dekker, 1994, pp 119–133 mediates endocytosis of low density lipoproteins via interaction 2. Blomhoff R, Green MH, Green JB, Berg T, Norum KR: Vitamin with apolipoprotein B100. J Biol Chem 270: 19417–19421, 1995 A metabolism: New perspectives on absorption, transport, and 19. Cui S, Verroust PJ, Moestrup SK, Christensen EI: Megalin/ storage. Physiol Rev 71: 951–990, 1991 gp330 mediates uptake of albumin in renal proximal tubule. Am J 3. Bang LE, Holm J, Svendsen TL: Retinol-binding protein and Physiol 271: F900–F907, 1996 in urine: New markers of renal function in essential 20. Moestrup SK, Cui S, Vorum H, Bregengård C, Bjørn SE, Norris hypertension and white coat hypertension? Am J Hypertens 9: K, Gliemann J, Christensen EI: Evidence that epithelial glycop- 1024–1028, 1996 rotein 330/megalin mediates uptake of polybasic drugs [see 4. Kabanda A, Vandercam B, Bernard A, Lauwerys R, van- comments]. J Clin Invest 96: 1404–1413, 1995 Ypersele-de SC: Low molecular weight proteinuria in HIV- 21. Christensen EI, Birn H, Verroust P, Moestrup SK: Membrane infected patients. Am J Kidney Dis 27: 803–808, 1996 receptors for endocytosis in the renal proximal tubule. Int Rev 5. Kato M, Kato K, Goodman DS: Immunocytochemical studies on Cytol 180: 237–284, 1998 the localization of plasma and of cellular retinol-binding proteins 22. Christensen EI, Gliemann J, Moestrup SK: Renal tubule gp330 is and of transthyretin (prealbumin) in rat liver and kidney. J Cell a calcium binding receptor for endocytic uptake of protein. Biol 98: 1696–1704, 1984 J Histochem Cytochem 40: 1481–1490, 1992 6. Usuda N, Kameko M, Kanai M, Nagata T: Immunocytochemical demonstration of retinol-binding protein in the lysosomes of the 23. Moestrup SK, Birn H, Fischer PB, Petersen CM, Verroust PJ, proximal tubules of the human kidney. Histochemistry 78: 487– Sim RB, Christensen EI, Nexø E: Megalin-mediated endocytosis 490, 1983 of transcobalamin-vitamin-B12 complexes suggests a role of the 7. Gerlach TH, Zile MH: Metabolism and secretion of retinol receptor in vitamin-B12 homeostasis. Proc Natl Acad Sci USA transport complex in acute renal failure. J Lipid Res 32: 515–520, 93: 8612–8617, 1996 1991 24. Willnow TE, Hilpert J, Armstrong SA, Rohlmann A, Hammer 8. Gerlach TH, Zile MH: Effect of retinoic acid and apo-RBP on RE, Burns DK, Herz J: Defective forebrain development in mice serum retinol concentration in acute renal failure. FASEB J 5: lacking gp330/megalin. Proc Natl Acad Sci USA 93: 8460–8464, 86–92, 1991 1996 9. Makover A, Soprano DR, Wyatt ML, Goodman DS: Localization 25. Gundersen TE, Lundanes E, Blomhoff R: Quantitative high- of retinol-binding protein messenger RNA in the rat kidney and performance liquid chromatographic determination of retinoids in perinephric fat tissue. J Lipid Res 30: 171–180, 1989 in human serum using on-line solid-phase extraction and column 10. Liu W, Carling T, Juhlin C, Rastad J, Ridefelt P, Åkerstro¨m G, switching: Determination of 9-cis-retinoic acid, 13-cis-retinoic Hellman P: Regulation of gp330/megalin expression by vitamins acid, all-trans-retinoic acid, 4-oxo-all-trans-retinoic acid and A and D. Eur J Clin Invest 28: 100–107, 1998 4-oxo-13-cis-retinoic acid. J Chromatogr B Biomed Sci Appl 11. Kerjaschki D, Noronha-Blob L, Sacktor B, Farquhar MG: Mi- 691: 43–58, 1997 crodomains of distinctive glycoprotein composition in the kidney 26. Christensen EI, Nielsen S, Moestrup SK, Borre C, Maunsbach proximal tubule brush border. J Cell Biol 98: 1505–1513, 1984 AB, de HE, Ronco P, Hammond TG, Verroust P: Segmental 12. Raychowdhury R, Niles JL, McCluskey RT, Smith JA: Autoim- distribution of the endocytosis receptor gp330 in renal proximal mune target in Heymann nephritis is a glycoprotein with homol- tubules. Eur J Cell Biol 66: 349–364, 1995 ogy to the LDL receptor. Science 244: 1163–1165, 1989 27. Birn H, Verroust PJ, Nexø E, Hager H, Jacobsen C, Christensen J Am Soc Nephrol 10: 685–695, 1999 Megalin-Mediated Endocytosis of RBP 695

EI, Moestrup SK: Characterization of an epithelial approximately (gp330) is a resident of coated pits of several absorptive epithelia 460-kD protein that facilitates endocytosis of - [Abstract]. J Cell Biol 98: 178, 1984 vitamin B12 and binds receptor-associated protein. J Biol Chem 36. Chatelet F, Brianti E, Ronco P, Roland J, Verroust P: Ultrastruc- 272: 26497–26504, 1997 tural localization by monoclonal antibodies of brush border an- 28. Le Panse S, Verroust P, Christensen EI: Internalization and tigens expressed by glomeruli. II. Extrarenal distribution. Am J recycling of glycoprotein 280 in BN/MSV yolk sac epithelial Pathol 122: 512–519, 1986 cells: A model system of relevance to receptor-mediated endo- 37. Soprano DR, Soprano KJ, Wyatt ML, Goodman DS: Induction of cytosis in the renal proximal tubule. Exp Nephrol 5: 375–383, the expression of retinol-binding protein and transthyretin in F9 1997 embryonal carcinoma cells differentiated to embryoid bodies. 29. Goodman DS: Plasma retinol-binding protein. In: The Retinoids, J Biol Chem 263: 17897–17900, 1988 edited by Sporn MB, Roberts AB, Goodman DS, Orlando, Ac- 38. Soprano DR, Wyatt ML, Dixon JL, Soprano KJ, Goodman DS: ademic Press, 1984, pp 41–88 Retinol-binding protein synthesis and secretion by the rat vis- 30. Bernard A, Lauwerys RR: Proteinuria: Changes and mechanisms ceral yolk sac: Effect of retinol status. J Biol Chem 263: 2934– in toxic nephropathies. Crit Rev Toxicol 21: 373–405, 1991 2938, 1988 31. Soprano DR, Soprano KJ, Goodman DS: Retinol-binding protein 39. Soprano DR, Smith JE, Goodman DS: Effect of retinol status on messenger RNA levels in the liver and in extrahepatic tissues of retinol-binding protein biosynthesis rate and translatable messen- the rat. J Lipid Res 27: 166–171, 1986 ger RNA level in rat liver. J Biol Chem 257: 7693–7697, 1982 32. Soprano DR, Soprano KJ, Goodman DS: Retinol-binding protein 40. Nagy NE, Holven KB, Roos N, Senoo H, Kojima N, Norum KR, and transthyretin mRNA levels in visceral yolk sac and liver Blomhoff R: Storage of vitamin A in extrahepatic stellate cells in during fetal development in the rat. Proc Natl Acad Sci USA 83: normal rats. J Lipid Res 38: 645–658, 1997 7330–7334, 1986 41. Le Panse S, Galceran M, Pontillon F, Lelongt B, van de Putte M, 33. Bavik C, Ward SJ, Chambon P: Developmental abnormalities in Ronco PM, Verroust PJ: Immunofunctional properties of a yolk cultured mouse embryos deprived of retinoic by inhibition of sac epithelial cell line expressing two proteins gp280 and gp330 yolk-sac retinol binding protein synthesis. Proc Natl Acad Sci of the intermicrovillar area of proximal tubule cells: Inhibition of USA 93: 3110–3114, 1996 endocytosis by the specific antibodies. Eur J Cell Biol 67: 34. Johansson S, Gustafson AL, Donovan M, Romert A, Eriksson U, 120–129, 1995 Dencker L: Retinoid binding proteins in mouse yolk sac and 42. Bavik CO, Eriksson U, Allen RA, Peterson PA: Identification chorio-allantoic placentas. Anat Embryol Berl 195: 483–490, and partial characterization of a retinal pigment epithelial mem- 1997 brane receptor for plasma retinol-binding protein. J Biol Chem 35. Doxsey S, Kerjaschki D, Farquhar MG: A large glycoprotein 266: 14978–14985, 1991