ARTICLES J Am Soc Nephrol 11: 2167–2178, 2000

Characterization of ANKRA, a Novel Repeat that Interacts with the Cytoplasmic Domain of Megalin

KATHERINE RADER,* ROBERT A. ORLANDO,† XIAOJING LOU,* AND MARILYN GIST FARQUHAR*† Departments of *Cellular and Molecular Medicine and †Pathology, University of California San Diego, La Jolla, California.

Abstract. Ankyrin-repeat family A protein (ANKRA) is a unique. By cell fractionation, ANKRA is found both in the novel protein that interacts directly and specifically with the cytosol and associated with membranes enriched in megalin in cytoplasmic tail of megalin in the two-hybrid system and L2 cells and proximal tubule cells. By immunofluorescence, glutathione-S-transferase pull-down assays. ANKRA has three ANKRA is concentrated near megalin along the plasma mem- ankyrin repeats and shows 61% overall homology to regulatory brane of L2 cells and in the kidney cortex is expressed in factor X, ankyrin repeat-containing protein. Mapping studies glomerular and proximal tubule epithelia which also express show that the three ankyrin repeats and C-terminus of ANKRA megalin. These observations suggest that ANKRA may play a are required for binding to a unique juxtamembrane, 19-amino unique role in megalin’s function as a clearance receptor in the acid sequence on the megalin tail. Point mutational analysis kidney and L2 cells. In addition, ANKRA may have other reveals that a proline-rich motif (PXXPXXP) within this region partners because northern blot analysis reveals that ANKRA is is the site of ANKRA binding. ANKRA interacts with megalin more broadly expressed than megalin, and by immunofluores- but not with low-density lipoprotein receptor related protein, in cence ANKRA is also expressed in connecting tubule cells and keeping with the fact that the sequence of the megalin tail is principal cells of collecting ducts.

Megalin (approximately 600 kD), a member of the low-density families, plasminogen, lactoferrin, thyroglobulin, Ca2ϩ, and lipoprotein receptor (LDLR) superfamily, is found in clathrin- receptor-associated protein (RAP), a chaperone for members of coated vesicles in glomerular and proximal tubule epithelial the LDLR gene family (1,3). cells of the kidney (1). Recent studies show that megalin is the Although considerable work has been done on ligand bind- main endocytic receptor in the proximal tubule and specifically ing to the extracellular domain of megalin, the role of its binds and internalizes a number of that are filtered by cytoplasmic domain in regulating megalin’s endocytic or exo- the glomerulus (2,3). These proteins include insulin (2); apo- cytic trafficking pathways (1,6) or putative Ca2ϩ-sensing func- lipoprotein H/I complexes; carriers for vitamins D3,B12, and A tion (7) is completely unknown. Megalin has a unique cyto- (retinol); and parathyroid hormone, among others (3,4). Mega- plasmic tail that shows little sequence similarity to other LDLR lin knockout mice exhibit proximal tubule reabsorption defi- family members except for conserved NPXY motifs (8). The ciency and a significant reduction in the number and size of fact that the cytoplasmic tail of receptors regulates their traf- organelles (clathrin-coated pits, endosomes, dense apical tu- ficking and signaling suggests that megalin has trafficking bules, and lysosomes) associated with endocytosis in the prox- itineraries or intracellular signaling pathways distinct from imal tubule (3,4). Thus, megalin expression is directly corre- those of other LDLR family members. The presence of lated with endocytic activity in the proximal tubule. In other dileucine, YXXØ, PXXP, and (S/T)XV motifs as well as systems, megalin has been shown to bind a number of other NPXY motifs, provides clues to the function of the megalin tail ligands from the lipoprotein and protease/protease inhibitor (7,8). Tyrosine-based motifs (NPXY, YXXØ) and dileucine motifs serve to direct receptors to clathrin-coated pits for rapid internalization or sorting within the trans-Golgi network by binding selectively to adaptor proteins AP-1, AP-2, or AP-3 Received February 22, 2000. Accepted May 30, 2000. Data deposition: The sequences reported in this article have been deposited in (9–11). This is in keeping with the fact that megalin, like other the GenBank database (Accession nos. hANKRA, AF314032; mANKRA, members of the LDL family, is internalized via clathrin-coated AF314031). pits (1,6,12). PXXP motifs bind to SH3 domains, (S/T)XV to Correspondence to Dr. Marilyn Gist Farquhar, Department of Cellular and PDZ domains, NPXY to PTB domains, and YXXØ to SH2 Molecular Medicine, University of California San Diego, 9500 Gilman Drive, domains (13). These interactions have been shown to be im- La Jolla, CA 92093-0651. Phone: 858-534-7711; Fax: 858-534-8549; E-mail: [email protected] portant for subcellular localization (14,15), endocytic traffick- 1046-6673/1112-2167 ing (16–19), and signaling (13–15,20) of some transmembrane Journal of the American Society of Nephrology proteins. However, nothing is known concerning the functions Copyright © 2000 by the American Society of Nephrology or molecular interactions of these motifs in the case of megalin. 2168 Journal of the American Society of Nephrology J Am Soc Nephrol 11: 2167–2178, 2000

To gain insight into the interactions of the megalin tail and in 10% fetal calf serum (Life Technologies, BRL, Grand Island, NY), thereby into megalin’s functions, we used the two-hybrid sys- MEM-alpha medium (Irvine Scientific, Santa Ana, CA). tem to identify proteins that bind to the cytoplasmic domain of megalin and therefore might be involved in directing megalin Library Screening and DNA Sequencing traffic or signaling. In this article, we report the initial charac- The cDNA encoding the cytoplasmic domain of rat megalin was terization and localization of a novel protein, ANKRA, which obtained by PCR from clone 217 (8) and cloned into the yeast specifically binds to the cytoplasmic domain of megalin. two-hybrid pGBT9 “bait” vector (Clontech). To isolate proteins that interact with the megalin tail, we screened a mouse kidney oligo-dT and random-primed MATCHMAKER cDNA library (Clontech) using Materials and Methods the yeast two-hybrid system as described previously (27,28). Of 5 cDNA and million transformants, three scored positive (blue) during lacZ selec- Two incomplete expressed sequence tags (ESTs) of human tion and were found to be identical by restriction digest grouping and ANKRA were obtained from the I.M.A.G.E. Consortium (Washing- sequencing the 5' ends. The insert contained the C-terminal half of ton University-Merck EST project, L. Hillier, unpublished data) and ANKRA (bp 745 to 1774) encoding aa138 to 312, the 3'UTR, and the sequenced by automated DNA sequencing. A complete human cDNA poly-A tail. To obtain the full-length cDNA, a mouse pituitary random sequence (1507 bp long) including the poly A tail was assembled from primed ␭gt-11 cDNA library (custom made by Stratagene and ob- overlapping human ESTs (GenBank accession numbers aa418089, tained from G. M. Rosenfeld, University of California San Diego) was aa442702, N6431, and AW079850). A plasmid containing the entire screened with a cDNA probe encoding aa138 to 312 of ANKRA sequence of human low-density lipoprotein receptor related protein according to the library manufacturer’s instructions. Three clones (LRP) was obtained from Joachim Herz (University of Texas were obtained: clone 11 encoded bp 1 to 1646 of ANKRA including Southwestern). the 5'UTR and 3'UTR, and clones 2 (full-length) and 3 (partial) were Rabbit megalin antiserum (459) was raised against a peptide from alternatively spliced in the C-terminus and 3'UTR, respectively. the cytoplasmic tail of megalin as described (21). Mouse monoclonal Clones were sequenced by either the dideoxynucleotide chain-termi- (mAb) anti-megalin IgG 20B was prepared as described previously nation method with Sequenase (United States Biochemical, Cleve- ␤ (22). Polyclonal anti- 1 integrin (130) was a generous gift from land, OH) or automated sequencing (DNA Sequencing Facility, Richard O. Hynes (MIT). The following antibodies were purchased: Scripps Research Institute, San Diego, CA). anti-HA mAb (BAbCO), goat anti-rabbit or goat anti-mouse IgG conjugated to horseradish peroxidase (HRP; Pierce, Rockford, IL), Sequence Analysis and Northern and cross-absorbed goat anti-rabbit Alexa 488 or anti-mouse Alexa Blot Analysis 594 F(ab')2 (Molecular Probes, Eugene, OR). Antisera were raised in rabbits against purified histidine (His)- On-line BLAST searches were performed via the National Center tagged full-length ANKRA and were affinity purified by coupling for Biotechnology Information at the National Institutes of Health purified His-tagged ANKRA (1-181) to AffiGel 15 beads (BioRad, (Bethesda, MD) (27). Scans for biologically significant sites, patterns, Hercules, CA) according to the manufacturer’s instructions. Bound and localization motifs on ANKRA were performed on-line using IgG was eluted with 0.1 M glycine (pH 2.5). By immunoblotting, 0.6 PROSITE (Swiss-Institute of Bioinformatics) and PSORT (Institute ␮g/ml affinity-purified anti-ANKRA (1-181) IgG detected 10 ng for Molecular and Cellular Biology, Osaka University, Osaka, Japan). purified His-tagged ANKRA. Sequence alignment was done using Clustal W and MacBoxShade ([email protected]) software programs. A Northern blot from eight murine tissues containing 2 ␮g poly Expression and Purification of Fusion Proteins Aϩ RNA per lane was purchased from Clontech and hybridized with Sequences of mouse ANKRA were subcloned by PCR into 32P-labeled probes prepared as described (28). pET28a(ϩ) vector (Novagen, Madison, WI). His-tagged fusion pro- teins were expressed in BL21(DE3) (Stratagene, La Jolla, GST-Precipitation Assays from Cell Lysates CA). Because the fusion proteins were insoluble in 1% Tween-20, Mouse ANKRA was subcloned by PCR behind an HA-epitope in they were solubilized from inclusion bodies with 8 M urea (in 20 mM the mammalian expression vector pcDNA3 (Invitrogen, Carlsbad, Tris [pH 7.8], 0.5 M NaCl, and 5 mM Imidazole), bound to Ni-NTA CA). For each assay, a 100-mm dish of HEK293 cells was transiently agarose beads (Qiagen, Valencia, CA), renatured on the beads, and transfected with 15 ␮g HA-tagged ANKRA in pcDNA3 by calcium eluted as described previously (23). phosphate precipitation. Forty-eight h posttransfection, cells were Glutathione-S-transferase receptor-associated protein (GST-RAP) solubilized in 10 mM CHAPS (3-[3(cholamidopropyl)-dimethylam- was prepared as described previously (24). Sequences of rat megalin monio-1-propanesulfonate] in buffer A [20 mM Hepes (pH 7.4), 150 tail (MT) or human LRP tail were subcloned by PCR into the mM NaCl, 2 mM CaCl ] containing protease inhibitors [0.5 mM pGEX-KG vector (Pharmacia, Piscataway, NJ). GST-tagged fusion 2 phenylmethylsulfonyl fluoride, 2 ␮g/ml aprotinin, 1 ␮g/ml chymosta- proteins were expressed in BL21 (pLys) or (DE3) bacteria (Strat- tin, 1 ␮g/ml leupeptin, 1 ␮g/ml antipain, 1 ␮g/ml pepstatin A (Sig- agene), solubilized in 1% Triton X-100 (in 50 mM Tris [pH 7.8], 0.4 ma)]) and centrifuged (14,000 rpm). Lysates were then incubated for M NaCl, 10% glycerol, 0.2 mM phenylmethylsulfonyl fluoride, 0.5 2 h at 4°C with 15 ␮g of purified GST-tagged fusion proteins ␮g/ml lysozyme), and purified on glutathione agarose beads (Phar- immobilized on glutathione beads. In some cases, the GST-MT fusion macia) according to the manufacturer’s instructions. protein was preincubated with purified (His)6-tagged ANKRA fusion protein, or the lysates were preincubated with anti-ANKRA(FL) se- Cell Culture rum or preimmune serum for1hat4°C. Beads were washed (4 to 5X) Rat L2 yolk sac cells were cultured as described previously (25,26). with lysis buffer, incubated (2 ϫ 5 min) with 20 ␮l of elution buffer Human embryonic kidney 293 (HEK293) cells, obtained from Dr. (20 mM glutathione ϩ 100 mM Tris-HCl [pH 8.0]), and boiled in 4X Joan Heller Brown (University of California San Diego), were grown Laemmli sample buffer. Eluted proteins were separated on 10% J Am Soc Nephrol 11: 2167–2178, 2000 ANKRA Interacts with the Megalin Tail 2169 sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS- inhibitors, and proteins from the supernatant and pellet were separated PAGE) gels and immunoblotted as described previously (6) using by 10% SDS gel electrophoresis. anti-HA mAb IgG (1:1000) followed by HRP-conjugated goat anti- Brush border membranes, basolateral membranes, and heavy en- rabbit IgG (1:3000). Bound antibodies were detected by chemilumi- dosome- and light endosome-enriched fractions were isolated from nescence using Super Signal (Pierce). the cortices of kidneys of Sprague-Dawley rats (Charles River Lab- oratories, Wilmington, MA) as described (31). Brush border (32,33) Interactions in the Yeast Two-Hybrid System and microvillar fractions (34,35) were prepared from the kidneys of Bait and prey constructs were co-transformed into yeast strain Sprague-Dawley rats or frozen rat kidneys by standard procedures. HF7c (Clontech), and one-to-one interaction was analyzed qualita- Brush border fractions consist of the entire apical region of proximal tively by a colony lift assay for ␤-gal using 5-bromo-4-chloro-3-idolyl tubule epithelia (33,36), whereas microvillar fractions consist of ve- (X-gal) (Clontech). For mapping interaction sites, deletion mutants of siculated microvilli, apical plasma membrane, and clathrin-coated pits the MT and ANKRA were subcloned into the pGBT9 “bait” vector (34,35). and pACT “prey” vectors (Clontech), respectively. Except for the mutants described below, clones were subcloned by PCR using MT- Immunocytochemical Localization of Endogenous or ANKRA-specific primers (sequences available upon request). ANKRA Complementary oligonucleotides containing point mutations were Kidneys of Sprague-Dawley rats were flushed with Cellgro DME annealed to form mutants MT(1-19), MT(LL/AA),and MT(PPP/ medium followed by perfusion with 4% paraformaldehyde (PFA) in AAA). The deletion mutants MT(1-58) and ANKRA(138-254) were 100 mM phosphate buffer (pH 7.4) for 15 min and immersion in 8% made by ligating a sequence with stop codons in all three frames into PFA for 1 h. L2 cells were fixed in 4% PFA for 15 min followed by blunt-ended NcoI sites present in MT(1-213) and ANKRA(138-312). 8% PFA for 45 min in phosphate buffer. Samples were cryoprotected All constructs were co-transformed with pGBT9 or pACT2 to show and frozen in liquid nitrogen (37). Semithin cryosections (0.5 to 1.0 that they did not activate transcription by themselves. ␮m) were cut on an Ultracut UCT microtome (Leica) equipped with an EM FCS cryoattachment at Ϫ100°C, placed on gelatin-coated In Vitro Pull-Down Assays microscope slides, and incubated for2hatroom temperature with In vitro transcription/translation of HA-tagged ANKRA in affinity-purified anti-ANKRA(1-181) IgG (0.06 ␮g/␮l) and/or anti- pCDNA3 was carried out using a TNT-T7-coupled reticulocyte lysate megalin mAb 20B (1:100), followed by incubation for 1 h with goat system (Promega, Madison, WI) in the presence of Amersham Redi- anti-rabbit Alexa 488 and/or anti-mouse Alexa 594 IgG as described vue-L [35S]methionine (Ͼ1000 Ci/mmol) according to the manufac- previously (6,29,30). Cells were observed with a Zeiss Axiophot turer’s instructions. Ten ␮g of GST-fusion proteins immobilized on microscope equipped for epifluorescence. Images were captured by a glutathione beads were incubated with 5 ␮l of reaction mixture Cohu camera and processed as TIFF files using SCION Image (NIH) containing 35S-labeled, in vitro translated HA-tagged ANKRA in 500 software. Images were colorized and superimposed using Adobe ␮l of 0.1% Tween-20/phosphate-buffered saline (pH 7.4) for2hat Photoshop 5.0 software to show overlap in the fluorescence pattern. 23°C with rotation. Beads were washed (3X) in the same buffer, Final images were printed on Kodak Ektatherm XLS paper. resuspended in 25 ␮l of Laemmli buffer, and boiled for 5 min. Proteins were separated on 12% SDS polyacrylamide gels and pro- Results cessed for autoradiography (Kodak Biomax film). Identification of Mouse and Human ANKRA To identify intracellular proteins that interact with megalin, Preparation and Analysis of Cell Fractions we used the cytoplasmic tail of megalin to screen a mouse Membrane (100,000 ϫ g pellet) and cytosolic (100,000 ϫ g su- kidney cDNA library in the yeast two-hybrid system. We pernatant) fractions were prepared from postnuclear supernatants obtained an interacting clone that encoded the C-terminal half (PNS) of rat kidney cortex, L2 cells, or HEK293 cells transfected with of a novel protein, which we named ANKRA for ankyrin- HA-tagged full-length mANKRA as described previously (28,29). To repeat protein, family A. To obtain the full-length mouse prepare a PNS from kidneys, the cortical region was trimmed from ANKRA sequence, we screened a mouse pituitary cDNAs frozen kidneys (obtained from Pelfreeze Biologicals, Rogers, AK), minced, homogenized in buffer A containing protease inhibitors using library and identified overlapping and full-length cDNAs. Fig- a Tissumizer (Tekmar Co.) for 3 min, and centrifuged twice at 600 ϫ ure 1A depicts the 312-aa sequence of ANKRA deduced from g for 10 min. To prepare a PNS from cultured cells, cells were its nucleotide sequence, which predicts a 33-kD protein. homogenized in ice-cold Tris-buffered saline containing protease in- ANKRA has three consecutive ankyrin repeats, two regions hibitors and centrifuged at 600 ϫ g for 5 min. The resultant PNS were with low homology to ankyrin repeats, four putative casein centrifuged at 100,000 ϫ g (29,30). Proteins in the pellet and super- kinase II phosphorylation sites, and two putative protein kinase natant were separated by 12% or 6% SDS-PAGE and immunoblotted C phosphorylation sites. The ankyrin repeat is a common 33-aa with affinity-purified anti-ANKRA(1-181) IgG (approximately 1 ␮g/ motif, originally identified in the cytoskeletal protein ankyrin, ml), anti-HA IgG (1:1000), or anti-MT(459) serum (1:2000), followed that is implicated in protein-protein interactions (38). Sequence by HRP-conjugated goat anti-rabbit or anti-mouse IgG (1:3000). After analysis indicates that ANKRA lacks putative membrane-span- enhanced chemiluminescence detection, protein bands were quanti- ning domains or targeting motifs, suggesting that it is most tated by densitometry using Alpha-imager software. Membranes were subjected to alkaline extraction as described likely a cytosolic or peripheral membrane protein. To obtain previously (29,30). Briefly, the 100,000 ϫ g pellet prepared from rat the amino acid sequence of human ANKRA, we assembled a full-length sequence from ESTs. Comparison of human and kidney cortex was resuspended in 0.2 M Na2CO3 (pH 11.5), incubated on ice for 30 min, and recentrifuged at 100,000 ϫ g. The resulting mouse ANKRA reveals 96% identity at the amino acid level pellet was lysed in 10 mM CHAPS in buffer A containing protease (Figure 1B). 2170 Journal of the American Society of Nephrology J Am Soc Nephrol 11: 2167–2178, 2000

Homology Between ANKRA and Regulatory Factor X Ankyrin Repeat-Containing Protein By searching protein sequence databases using BLAST, we found that ANKRA has significant homology to a protein known as regulatory factor X ankyrin repeat-containing protein (RFXANK) in all regions except the N-terminal domain (Fig- ure 1, B and C). RFXANK (33 kD) functions in the assembly of the RFX-5 transcription complex that upregulates expres- sion of MHC class II proteins (39,40). RFXANK is also a substrate of Raf-1, a serine/threonine kinase, and enhances Raf-1 activation by epidermal growth factor (41). Amino acids 126 to 308 of ANKRA are 66% identical to aa 67 to 260 in RFXANK (Figure 1, B and C). The unique N-terminal domain of ANKRA (aa 1 to 122) has a predicted ␣-helical/loop secondary structure in contrast to the N-terminus (aa 1 to 64) of RFXANK, which contains a PEST domain (Figure 1C) (40). Interestingly, the N-terminus of ANKRA contains one putative casein kinase II phosphoryla- tion site (thr 132) (Figure 1A), also found in RFXANK, which is known to be phosphorylated in vivo (40,41). The next closest protein to ANKRA is integrin-linked kinase, where the three ankyrin repeats in ANKRA are 30% identical to ankyrin repeats in integrin-linked kinase. The lack of any significant (Ͼ50%) sequence homology between ANKRA and any protein other than RFXANK suggests that ANKRA and RFXANK are members of a novel gene family.

ANKRA mRNA Transcripts Are Widely Expressed and Alternatively Spliced By Northern blotting, three transcripts were detected; a major mRNA transcript of approximately 2 kb was detected in all tissues tested: brain, heart, spleen (less abundant), lung, liver, , kidney, and testis. In addition, a tran- script of approximately 2.7 kb was detected in brain, spleen, lung, liver, and kidney, and a transcript of approximately 2.5 kb was detected in testis (Figure 2A). Because megalin is Figure 1. ANKRA is a novel protein. (A) The amino acid (aa) sequence expressed in epithelial cells of the kidney, lung, testis, and of ANKRA contains three ankyrin repeats (ank 1 to 3) as well as two brain (1,8), ANKRA and megalin mRNA are found together in adjoining regions with low homology to ankyrin repeats (ank). It also these tissues. However, ANKRA has a broader tissue expres- contains four putative casein kinase II (*) and two putative protein kinase sion than megalin. C(’) phosphorylation sites. A vertical line indicates a C-terminal alter- Detection of three mRNA transcripts by Northern blotting native splice site found in a library clone and in expressed sequence tags suggested that ANKRA is alternatively spliced. This was con- (EST; GenBank AC# AI060551 and 731339), and an arrow marks the beginning of the two-hybrid clone. (B) ANKRA has significant homol- firmed by library screening and searching EST databases, ogy to RFXANK. Identical amino acids are highlighted in black, and where we identified three alternatively spliced ANKRA homologous amino acids are highlighted in grey. mANKRA, mouse cDNAs: ANKRA(FL), ANKRA Long 3'UTR, and ANKRA ANKRA; hANKRA, human ANKRA; mRFXANK, mouse RFXANK; Spliced C-terminus (Figure 2B). ANKRA(FL) (1774 bp) cor- hRFXANK, human RFXANK. (C) Schematic diagrams of ANKRA and responds in size to the major mRNA of approximately 2 kb RFXANK showing the conserved ankyrin repeats (I to III) (f) and ank (Figure 2A), and ANKRA Long 3'UTR (approximately 2374 regions (1) as well as the nonconserved N-termini. The unique N- d ␣ bp) may correspond to the minor mRNA of approximately 2.5 terminal domain of ANKRA ( ) is predicted to form several -helixes ␤ or 2.7 kb (Figure 2A). ANKRA Spliced C-terminus is alterna- separated by loops and -turns, whereas RFXANK contains a PEST domain (s). The percentage of amino acid identity between the con- tively spliced at aa 295 and encodes a protein that is 8 aa served domains is indicated. shorter than ANKRA. J Am Soc Nephrol 11: 2167–2178, 2000 ANKRA Interacts with the Megalin Tail 2171

serum (Figure 3B, lane 4) did not. These results demonstrate the specificity of ANKRA binding to MT and confirm the two-hybrid results.

The Ankyrin-Repeat Domain and C-Terminus of ANKRA Are Required for Interaction with the MT To map the megalin binding site in ANKRA, we tested the ability of MT to interact with ANKRA deletion mutants in the yeast two-hybrid system. When truncation was carried out from the N-terminus of ANKRA (Figure 4A), we observed strong interaction of MT with constructs containing the ankyrin repeats and C-terminus (aa 138 to 312 and 177 to 312) but not with the C-terminal domain alone (aa 254 to 312). When truncation was carried out from the C-terminus, we found that ANKRA (aa 138 to 254) containing ankyrin repeats 1 and 2 did not bind to MT. Moreover, when the extreme C-terminus (aa 290 to 312) was deleted, the N-terminal domain and ankyrin repeats (aa 1 to 290) were not capable of interacting with MT. Together, these results suggest that both the ankyrin repeats and the last 22 aa of ANKRA, which encompass the ank domain and the alternatively spliced C-terminus, are necessary for interaction with the MT.

The N-Terminal Domain of the MT (aa 1 to 19) Binds to ANKRA and Contains a Proline-Rich Motif PXXPXXP Necessary for the Interaction Next we mapped the MT binding site for ANKRA in the two-hybrid assay. We carried out truncation from the C-termi- nus of MT (Figure 4B) and narrowed the binding site to aa 1 to 58. The region C-terminal of this binding site, aa 58 to 213, Figure 2. Tissue distribution and alternative splicing of ANKRA mRNA did not interact with ANKRA, demonstrating that NPXY mo- transcripts. (A) Upper panel: Northern blot of poly(A)ϩmRNA from tifs in this region of MT are not required for the interaction. adult mouse tissues probed with a mouse ANKRA (aa 138 to 312) cDNA probe. Lower panel: The same northern blot probed for human ␤-actin. Next we generated a smaller mutant, MT(1 to 19), which (B) Schematic representation of alternatively spliced ANKRA cDNAs. contains a dileucine and a proline-rich motif, and found that it (1) Mouse cDNA encoding full-length ANKRA. (2) Mouse cDNA also bound to ANKRA. To determine whether the dileucine or containing a longer 3'UTR of approximately 600 bp. (3) Shorter mouse proline-rich motifs are required for the interaction of MT with cDNA alternatively spliced on the C-terminus and 3'UTR (pituitary ANKRA, we mutated the dileucine motif to alanines MT(LL/ library clone 2, ESTs: GenBank AC# AI060551, Z31339). The last AA) or the prolines P10XXP13XXP16 to alanines MT(PPP/ cDNA is incomplete as it lacks the poly(A) tail (pA). AAA) and found that ANKRA bound to MT(LL/AA) but did not interact with MT(PPP/AAA). From these data, we con- clude that the proline but not the dileucine motif is required for ANKRA Interacts Specifically with the MT in binding of MT to ANKRA. Cell Lysates To verify the interaction between ANKRA and MT (1 to To verify the interaction between MT and ANKRA, we 19), we deleted aa 1 to 20 of MT and tested its binding to performed precipitation assays on cell lysates prepared from ANKRA using in vitro pull-down assays. Figure 4C shows that cells transfected with HA-tagged ANKRA with MT immobi- 35S-labeled, HA-tagged ANKRA bound to full-length MT lized on glutathione beads (GST-MT). We found that ANKRA (lane 2) but did not bind to MT(20 to 213) (lane 3) or GST bound to GST-MT (Figure 3A, lane 2) but not to GST alone alone (lane 1). Thus, both in the yeast two-hybrid system and (Figure 3A, lane 1). When GST-MT was preincubated with by in vitro pull-down assays, we found that aa 1 to 19 of MT equivalent or 10-fold amounts of recombinant His-tagged are required for interaction with ANKRA. ANKRA before incubation with cell lysate, binding of HA- tagged ANKRA was reduced (Figure 3A, lanes 3 and 4). ANKRA Binds to Megalin but not to LRP Preincubation with less than equivalent amounts of His-tagged We also used in vitro GST-pull-down assays to determine ANKRA (Figure 3A, lanes 5 and 6) had little or no effect. whether ANKRA binds to the cytoplasmic tail of LRP. LRP Moreover, preincubating cell lysates with anti-ANKRA(FL) has little sequence homology to the MT except for NPXY polyclonal antiserum abolished binding to GST-MT (Figure motifs (8) and therefore would not be expected to bind to 3B, lane 3), whereas an equivalent amount of preimmune ANKRA. HA-tagged ANKRA protein (44 kD) did not bind to 2172 Journal of the American Society of Nephrology J Am Soc Nephrol 11: 2167–2178, 2000

Figure 3. ANKRA interacts specifically with the GST-MT in GST precipitation assays. (A) Lysates from HEK 293 cells overexpressing HA-tagged ANKRA were incubated with GST (lane 1) or GST-MT immobilized on glutathione-agarose beads (lanes 2 to 6) in the pres- ence or absence of varying concentrations of purified His-tagged ANKRA for 2 h. Interacting proteins were separated by SDS-PAGE, and ANKRA was detected by Western blotting using anti-HA mAb. HA-tagged ANKRA binds to GST-MT (lane 2) but not to GST alone (lane 1). Addition of 80 ␮g (lane 3) or 8 ␮g (lane 4) of purified His-tagged ANKRA to GST-MT eliminates interaction, whereas ad- dition of 0.8 or 0.08 ␮g (lanes 5 and 6) has minimal or no effect. Lane 7: 3% of total cell lysate. (B) Cell lysates prepared as in A were preincubated with ANKRA immune or preimmune antisera (1 h), Figure 4. Mapping the interacting domains on ANKRA and the MT using followed by incubation with GST (lane 1) or GST-MT (lanes 2 to 4) the yeast two-hybrid system. (A) Ankyrin repeats (I to III) and the C-terminus for2hasinA.HA-tagged ANKRA binds to GST-MT (lane 2) but not (aa 290 to 312) of ANKRA are required for interaction with MT in the yeast ␮ to GST alone (lane 1). Addition of 2.5 l of anti-ANKRA(FL) serum two-hybrid system. MT interacts strongly with full-length ANKRA, (lane 3) but not preimmune serum (lane 4) eliminates binding to ANKRA (138 to 312), the original two- hybrid clone, and ANKRA (177 to GST-MT. 312)—all of which contain ankyrin repeats (I to III) and the C-terminal domain. MT does not interact with the C-terminal domain alone (ANKRA 254 to 312), ankyrin repeats (I to II) (ANKRA 138 to 254), or the N-terminal GST-LRP tail (Figure 4C, lane 4) or GST-RAP (binds to the domain plus ankyrin repeats (ANKRA 1 to 290). (B) The N-terminal domain ectodomain of megalin and LRP) used as a negative control of MT (aa 1 to 19) is required for interaction with ANKRA in the yeast (Figure 4C, lane 5). These results indicate that ANKRA binds two-hybrid system. ANKRA interacts strongly with MT(1 to 172), MT(1 to specifically to megalin but not to LRP. We were unable to 58), and the short sequence MT(1 to 19). Deletion of aa 1 to 58, MT(58 to 213), abolishes interaction with ANKRA. Point mutational analysis of MT(1 assess the binding between the LRP tail and ANKRA using the to 19) shows that ANKRA interacts with MT(LL/AA) but not MT(PPP/ yeast two-hybrid system because, as previously reported (42), AAA), demonstrating that the proline-rich motif (PXXPXXP) but not the the LRP-tail bait is weakly self-activating. dileucine motif (LL) is necessary for the interaction. (Note: aa 1 to 213 of the MT correspond to aa 4423 to 4635 of megalin.) The ␤-gal filter assay was ANKRA Is Found in Both Cytosolic and Membrane performed on (Leu-, Trp-) plates and the color intensity was qualitatively Fractions of Rat Kidney Cortex and L2 Cells scored at 4 to7hasthefollowing: Ϫ, no color; ϩ, weak color; ϩϩ, Next we investigated whether endogenous ANKRA is asso- intermediate color; ϩϩϩ, strong color; ϩϩϩϩ, strongest color. ciated with membrane or cytosolic fractions in rat kidney (C) Deleting aa 1 to 20 abolishes the interaction with MT in in vitro cortex and L2 yolk sac cells. We used L2 cells because the pull-down assays. Radiolabeled, in vitro translated HA-tagged distribution of megalin in these cells has been well character- ANKRA binds to GST-MT (lane 2) but not to GST-MT(20 to 213) ized (1,25,26). First, we carried out immunoblotting on cyto- (lane 3) or GST alone (lane 1). ANKRA also does not bind GST-LRP tail (lane 4) or GST-RAP, a negative control (lane 5). Lane 6: 10% of solic (100,000 g supernatant) or membrane (100,000 g pellet) the total in vitro-translated product. Equivalent amounts of GST fractions using anti-ANKRA(1 to 181) IgG that had been fusion proteins were bound to glutathione beads and incubated with affinity purified on the unique N-terminus of ANKRA to 35S-labeled, in vitro-translated HA-tagged ANKRA for 2 h. Bound eliminate antibodies that might cross react with ankyrin-repeats proteins were separated by SDS-PAGE and detected by of other proteins. In fractions of rat kidney cortex, we detected autoradiography. two bands, 46 and 44 kD, which presumably represent post- J Am Soc Nephrol 11: 2167–2178, 2000 ANKRA Interacts with the Megalin Tail 2173

Figure 5. ANKRA is found in both cytosolic (100,000 g supernatant) and membrane (100,000 g pellet) fractions from kidney cortex and L2 cells. (A) Cytosolic (C) and membrane (M) fractions prepared from a postnuclear supernatant (PNS) of rat kidney cortex were immunoblotted with approximately 1 ␮g/ml affinity-purified anti-ANKRA(1 to 181) IgG (lower panel) or anti-megalin 459 serum (upper panel). Two protein bands are detected—a 46-kD band found in both cytosolic and membrane fractions and a 44-kD band found exclusively in the cytosolic fraction. (B) In L2 cells, a single protein of approximately 46 kD is seen in both C and M fractions. Megalin used as a positive control is found only in the membrane fraction. (C) ANKRA is a peripheral membrane protein. M fractions prepared from rat kidney cortex were treated with 0.2 ϫ MNa2CO3 (pH 11.5) and centrifuged at 100,000 g. ANKRA is released into the supernatant (S) and is not seen in the pellet (P), whereas megalin remains associated with the membrane pellet (P) after alkaline treatment. Fractions were immunoblotted as in A.

translationally modified and/or alternatively spliced forms of Localization of Endogenous ANKRA in L2 Cells and ANKRA (Figure 5A). The 46-kD band was found in both Rat Kidney cytosolic (40%) and membrane (60%) fractions, whereas the Next we carried out double labeling by immunofluorescence 44-kD band was found exclusively in the cytosol. In L2 cells, on semithin cryosections of L2 cells using affinity-purified we detected a single, 46-kD band in both cytosolic (80%) and anti-ANKRA(1 to 181) IgG (Figure 7A) and anti-megalin mAb membrane (20%) fractions (Figure 5B). These results show 20B (Figure 7B). ANKRA was found in the cytoplasm con- that the number of bands and the percentage of ANKRA found centrated predominantly along the plasma membrane (Figure in the cytosolic and membrane pools vary depending on the 7B), where it partially overlapped with megalin (Figure 7C). source. In rat kidney cortex, we detected ANKRA in glomeruli To determine the nature of ANKRA’s association with (Figure 8, A and C), proximal tubules (Figure 8, A and B), membranes, we carried out alkaline extraction (pH 11.5) on connecting tubules (Figure 8, A and D), and collecting ducts (data not shown). In some proximal tubule cells, ANKRA was membrane fractions from rat kidney. We found that ANKRA concentrated near the brush border where megalin is located was released into the supernatant after alkaline treatment (Fig- (Figure 8B); in others, ANKRA was distributed throughout the ure 5C), indicating that ANKRA is peripherally associated cytoplasm. ANKRA was also distributed throughout the cyto- with membranes and is not an integral membrane protein. plasm in glomerular epithelial cells (Figure 8C), which express megalin, and in selected cells of connecting tubules (Figure 8D) and in principal cells of collecting ducts. ANKRA Is Found in Apical Membrane Fractions of Proximal Tubules Discussion Megalin is found in glomerular epithelia and in greatest To date, no cytoplasmic proteins that bind to megalin have abundance on the apical domain of proximal tubule epithelia been identified. Because the cytoplasmic tail of megalin has (35,43). To determine whether ANKRA co-distributes with putative tyrosine and dileucine-based sorting signals (10,11) as megalin in proximal tubules, we analyzed fractions enriched in well as putative motifs for binding SH3, SH2, and PDZ do- brush border membranes and basolateral membranes (31). By mains (13), it can be anticipated to bind multiple intracellular immunoblotting, we detected ANKRA in fractions enriched in proteins. Using the cytoplasmic domain of megalin as bait in light and heavy endosomes (data not shown) and brush border the yeast two-hybrid system, we screened a mouse kidney membranes but not basolateral membranes (Figure 6A). We cDNA library and identified a novel 312 aa, mammalian pro- also prepared purified brush border (32) and microvillar frac- tein, ANKRA, which has three ankyrin repeats and is highly tions (34) from rat kidney cortex and detected ANKRA in both conserved (approximately 96%) in mouse and human. fractions by immunoblotting (Figure 6B). Collectively, these ANKRA shows 61% overall homology to another ankyrin results suggest that ANKRA is enriched in apical membranes repeat–containing protein, RFXANK, indicating that ANKRA of the proximal tubule. and RFXANK are members of a novel gene family. Ankyrin 2174 Journal of the American Society of Nephrology J Am Soc Nephrol 11: 2167–2178, 2000

Figure 6. ANKRA is enriched in brush border and microvillar (MV) fractions. (A) Fractions of brush border membranes (BBM) and basolateral membranes (BL) were prepared from fluorescein dextran-loaded kidneys by magnesium precipitation according to the method of Hammond et al. (31). ANKRA is enriched in BBM but not BL. Megalin is found both in the BBM ␤ fraction and in the BL fraction. 1-integrin, a basolateral membrane protein, is enriched in the BL fraction. (B) ANKRA and megalin are detected in purified BB and MV fractions. The 34-kD band in the MV fraction is believed to be a degradation product of ANKRA. Brush border (BB) frac- tions were prepared from kidney homogenate (H) according to Thuneberg and Rostgaard (32). MV fractions were prepared from kidney by calcium precipitation (34). Fractions were immunoblotted with approximately 1 ␮ ␤ g/ml affinity-purified anti-ANKRA(1 to 181) IgG and anti- 1-integrin or anti-megalin 459 serum (1:2000). repeats are a common motif, originally identified in ankyrin, that have subsequently been found in more than 400 proteins with a variety of functions (44). Ankyrin, a cytoskeletal protein with 24 tandem ankyrin repeats, functions to link spectrin-actin networks to the cytoplasmic domains of transmembrane pro- teins (44), such as NaϩK-ATPase (45). Using GST in vitro pull-down and GST-precipitation assays on cell lysates, we demonstrated specific interaction between ANKRA and megalin in vitro and in cell lysates. Moreover, we have shown by immunofluorescence that ANKRA is present in cells known to express megalin, including rat L2 yolk sac cells and rat kidney proximal tubule and glomerular epithelial cells, where it is concentrated near megalin along the plasma mem- brane. There are two pools of ANKRA—one cytoplasmic and Figure 7. Immunofluorescence localization of ANKRA and megalin the other membrane associated. The membrane-associated pool in a semithin cryosection of L2 cells. (A) ANKRA is detected in the co-distributes with megalin in purified brush border and mi- cytoplasm and concentrated at the cell periphery along the plasma crovillar fractions and is peripherally associated with mem- membrane (PM). (B) Localization of megalin in the same section branes as shown by alkaline extraction. These collective results showing its characteristic distribution in a punctate pattern at the cell indicate that ANKRA interacts with megalin and probably surface reflecting its coated pit localization. Note that L2 cells express plays a role in megalin’s functions. In addition, ANKRA most variable levels of megalin (25). (C) When A and B are merged, the likely has interactions and functions in kidney epithelia and distribution of ANKRA and megalin partially overlap. Cells were other organs. This is suggested by our findings that ANKRA fixed in paraformaldehyde, and semithin cryosections were incubated can undergo alternative splicing, is more broadly expressed in with affinity-purified anti-ANKRA(1 to 181) IgG and anti-megalin tissues and cells than megalin, and is abundantly expressed in 20B mAb followed by incubation in cross-absorbed goat anti-rabbit Alexa 488 and anti-mouse Alexa 594 IgG. Bar, 10 ␮M. connecting and collecting tubules of the kidney cortex. J Am Soc Nephrol 11: 2167–2178, 2000 ANKRA Interacts with the Megalin Tail 2175

Figure 8. Immunofluorescence localization of ANKRA in rat kidney cortex. Phase-contrast (left panel) and immunofluorescence (right panel) of semithin sections fixed with paraformaldehyde and incubated with affinity-purified anti-ANKRA(1 to 181) IgG. (A, A') ANKRA is seen in the glomerulus (G), proximal tubule (PT), and connecting tubule (CNT). (B, B') ANKRA is concentrated in the apical region of a PT cell between the nucleus (N) and brush border (arrows). (C, C') High-magnification view of a portion of a glomerulus showing ANKRA distributed throughout the cytoplasm of glomerular epithelial cells (GEC). (D, D') CNT showing expression of ANKRA in connecting tubule cells. L, lumen; MV, microvilli; N, nucleus. Bar: 50 ␮minA,10␮m in B through D.

Mapping studies reveal that the three ankyrin repeats and the shown to depend on adjacent nonankyrin repeat regions as well C-terminus of ANKRA are required for optimal interaction as the number of stacked ankyrin repeats and the surface with MT. As ANKRA is alternatively spliced, incorporation of contact residues (44). It is interesting that the ankyrin repeat different C-termini into ANKRA could contribute to the bind- region that is highly conserved in both ANKRA and RFXANK ing specificity and/or the structural stability of ANKRA be- facilitates homodimerization of RFXANK and its binding to cause the binding specificity of ankyrin repeat regions has been Raf-1 kinase (41). Dimerization of ANKRA could allow si- 2176 Journal of the American Society of Nephrology J Am Soc Nephrol 11: 2167–2178, 2000 multaneous binding of ANKRA to megalin and other proteins Alternatively or simultaneously, ANKRA could act as a such as raf-1 kinase. Ankyrin repeat domains have the potential signaling molecule. ANKRA has significant homology to RFX- to bind multiple proteins simultaneously because they have ANK, which contributes to the assembly of the RFX-5 transcrip- three potential surface contacts (␤-hairpins, ankyrin groove, tion complex and is a substrate of raf-1, a cytosolic serine/ and outer surface of the ␣-helical bundle) (44). threonine kinase (39–41). It also potentiates raf-1 activation Mapping studies demonstrate that interaction between induced by epidermal growth factor (41). ANKRA may interact ANKRA and aa 1 to 19 of the MT requires a proline-rich motif with Raf-1 because the RFXANK binding domain for Raf-1 (PXXP), which may provide surface contacts or contribute to kinase (41) is highly conserved (77% homologous) in ANKRA. the conformation of the binding domain on megalin. Three Future studies will be aimed at identifying other interacting protein modules—SH3, WW, and EVH1 domains—are known partners of ANKRA and determining ANKRA’s role in mega- to recognize proline-rich consensus sequences, but only SH3 lin’s function. domains have been shown to interact with the PXXP consensus sequence (13). In this article, we identified a novel protein- Note added in proof: Two articles (Biochem J 347: 613– protein interaction between an ankyrin repeat domain and a 621, 2000, and J Biol Chem 275: 25616–25624, 2000) ap- proline-rich region. peared while this manuscript was in press, reporting additional ANKRA is highly homologous to RFXANK (alias Tvl-1 and cytosolic proteins that interact with the megalin tail. RFX-B) (39,40) between aa 122 and 308 (see Figure 1C), Acknowledgments which suggests that an early gene duplication gave rise to We thank Tammie McQuistan, Chris Hofeditz, and Mark Wahren- ANKRA and RFXANK. In addition, several genes, including brock for technical assistance, Dr. Akihiko Saito for the megalin tail ANKRA and RFXANK, are duplicated on chromosomal bands construct, Dr. Goeff Rosenfeld for providing the mouse pituitary 5q12–13 (46) and 19p12 (39), respectively. Although there is cDNA library, and Dr. Joachim Herz for providing the LRP construct. extensive sequence homology between the ankyrin repeat re- This work was supported by an American Heart Association, Califor- gions and C-termini of ANKRA and RFXANK, they have nia Affiliate Predoctoral Fellowship to K.R. and National Institutes of unique N-termini. ANKRA has a predicted ␣-helical/loop do- Health Research Grant DK17724 to M.G.F. K.R. is a graduate student main, whereas RFXANK has a PEST domain, a sequence that in the Biomedical Sciences Graduate Program at UCSD. directs ubiquination and subsequent degradation in proteo- somes (47). The N-terminal domain of ANKRA does not show References any homology to other mammalian proteins and most likely 1. Farquhar MG, Saito A, Kerjaschki D, Orlando RA: The Hey- has a unique function. mann nephritis antigenic complex: Megalin (gp330) and RAP. J Am Soc Nephrol 6: 35–47, 1995 What is the role of ANKRA in megalin’s function? Several 2. Orlando RA, Rader K, Authier F, Yamazaki H, Posner BI, possibilities come to mind. Ankyrin repeats commonly have a Bergeron JJ, Farquhar MG: Megalin is an endocytic receptor for scaffold function and are required for the interaction with MT. insulin. J Am Soc Nephrol 9: 1759–1766, 1998 This raises the possibility that ANKRA could facilitate endo- 3. Christensen EI, Willnow TE: Essential role of megalin in renal cytosis by linking megalin to components of the proximal tubule for vitamin homeostasis. J Am Soc Nephrol 10: or endocytic machinery. In yeast, ankyrin repeat-containing 2224–2236, 1999 protein 1 (Akr1p) interacts with the cytoplasmic tails of pher- 4. Christensen EI, Birn H, Verroust P, Moestrup SK: Membrane omone receptors and is required for their constitutive and/or receptors for endocytosis in the renal proximal tubule. Int Rev ligand-mediated endocytosis (48). It has been hypothesized Cytol 180: 237–284, 1998 that Akr1p together with other proteins (e.g., iqg1p) regulates 5. Osman MA, Cerione RA: Iqg1p, a yeast homologue of the actin organization during endocytosis (5). In addition, it was mammalian IQGAPs, mediates cdc42p effects on the actin cy- toskeleton. J Cell Biol 142: 443–455, 1998 recently discovered that the fourth ankyrin repeat of ankyrin R 6. Czekay RP, Orlando RA, Woodward L, Lundstrom M, Farquhar interacts with high affinity with the heavy chain of clathrin and MG: Endocytic trafficking of megalin/RAP complexes: Dissoci- plays a role in internalization of LDL (49). ation of the complexes in late endosomes. Mol Biol Cell 8: Although ANKRA does not contain an SH3 domain, it binds 517–532, 1997 to a proline-rich motif on MT that is a consensus sequence for 7. Hjälm G, Murray E, Crumley G, Harazim W, Lundgren S, the binding of SH3 proteins. Another possibility is that Onyango I, Ek B, Larsson M, Juhlin C, Hellman P, Davis H, ANKRA may affect megalin endocytosis by inhibiting binding Akerstrom G, Rask L, Morse B: Cloning and sequencing of ϩ of the MT to an SH3-containing protein. Recent studies have human gp330, a Ca2 -binding receptor with potential intracel- shown that removal of the SH3-domain binding sites from the lular signaling properties. Eur J Biochem 239: 132–137, 1996 third intracellular loop of the dopamine D4 receptor promotes 8. Saito A, Pietromonaco S, Loo AK, Farquhar MG: Complete receptor internalization (50). It has also been shown that the cloning and sequencing of rat gp330/“megalin,” a distinctive member of the low density lipoprotein receptor gene family. third intracellular loop of the ␤ -adrenergic receptor interacts 1 Proc Natl Acad Sci USA 91: 9725–9729, 1994 with the SH3 domain of endophilin 1 and that overexpression 9. Hirst J, Robinson MS: Clathrin and adaptors. Biochim Biophys of endophilin 1 enhances internalization of the receptor (16). Acta 1404: 173–193, 1998 The latter effect may be because the SH3 domain of endophilin 10. Kirchhausen T, Bonifacino JS, Riezman H: Linking cargo to binds to proteins of the endocytic machinery and functions in vesicle formation: Receptor tail interactions with coat proteins. late steps of clathrin-coated vesicle formation (16,17,50). Curr Opin Cell Biol 9: 488–495, 1997 J Am Soc Nephrol 11: 2167–2178, 2000 ANKRA Interacts with the Megalin Tail 2177

11. Le Borgne R, Hoflack B: Mechanisms of protein sorting and coat ing protein for Galphai heterotrimeric G proteins, is located on assembly: insights from the clathrin-coated vesicle pathway. clathrin-coated vesicles. Mol Biol Cell 9: 1123–1134, 1998 Curr Opin Cell Biol 10: 499–503, 1998 30. Lin P, Le-Niculescu H, Hofmeister R, McCaffery JM, Jin M, 12. Kerjaschki D, Miettinen A, Farquhar MG: Initial events in the Hennemann H, McQuistan H, DeVries L, Farquhar MG: The formation of immune deposits in passive Heymann nephritis: mammalian calcium-binding protein, nucleobindin (CALNUC), gp330-anti-gp330 immune complexes form in epithelial coated is a Golgi resident protein. J Cell Biol 141: 1515–1527, 1998 pits and rapidly become attached to the glomerular basement 31. Hammond TG, Verroust PJ, Majewski RR, Muse KE, Oberley membrane. J Exp Med 166: 109–128, 1987 TD: Heavy endosomes isolated from the rat renal cortex show 13. Pawson T, Scott JD: Signaling through scaffold, anchoring, and attributes of intermicrovillar clefts. Am J Physiol 267: F516– adaptor proteins. Science 278: 2075–2080, 1997 F527, 1994 14. Fanning AS, Anderson JM: Protein modules as organizers of 32. Thuneberg L, Rostgaard J: Isolation of brush border fragments membrane structure. Curr Opin Cell Biol 11: 432–439, 1999 from homogenates of rat and rabbit kidney cortex. Exp Cell Res 15. Margolis B, Borg JP, Straight S, Meyer D: The function of PTB 51: 123–140, 1968 domain proteins. Kidney Int 56: 1230–1237, 1999 33. Rodman J, Kerjaschki D, Merisko E, Farquhar M: Presence of an 16. Tang Y, Hu LA, Miller WE, Ringstad N, Hall RA, Pitcher JA, extensive clathrin coat on the apical plasmalemma of the rat DeCamilli P, Lefkowitz RT: Identification of the endophilins kidney proximal tubule cell. J Cell Biol 98: 1630–1636, 1984 (SH3p4/p8/p13) as novel binding partners for the beta1-adrener- 34. Kerjaschki D, Farquhar MG: The pathogenic antigen of Hey- gic receptor. Proc Natl Acad Sci USA 96: 12559–12564, 1999 mann nephritis is a membrane glycoprotein of the renal proximal 17. Simpson F, Hussain NK, Qualmann B, Kelly RB, Kay BK, tubule brush border. Proc Natl Acad Sci USA 79: 5557–5581, McPherson PS, Schmid SL: SH3-domain-containing proteins 1982 function at distinct steps in clathrin-coated vesicle formation. Nat 35. Kerjaschki D, Noronha-Blob L, Sacktor B, Farquhar MG: Micro- Cell Biol 1: 119-124, 1999 domains of distinctive glycoprotein composition in the kidney 18. Cao TT, Deacon HW, Reczek D, Bretscher A, von Zastrow M: A proximal tubule brush border. J Cell Biol 98: 1505–1513, 1984 kinase-regulated PDZ-domain interaction controls endocytic sorting 36. Rodman JS, Mooseker M, Farquhar MG: Cytoskeletal proteins of the ␤2-adrenergic receptor. Nature 401: 286–290, 1999 of the rat kidney proximal tubule brush border. Eur J Cell Biol 42: 319–327, 1986 19. Wang Z, Moran MF: Requirement for the adapter protein GRB2 37. McCaffery JM, Farquhar MG: Localization of GTPases by indi- in EGF receptor endocytosis. Science 272: 1935–1939, 1996 rect immunofluorescence and immuno-electron microscopy. 20. Willnow TE, Nykjaer A, Herz J: Lipoprotein receptors: New Methods Enzymol 257: 259–279, 1995 roles for ancient proteins. Nat Cell Biol 1: E157–E162, 1999 38. Bennett V: adaptors between diverse plasma mem- 21. Czekay RP, Orlando RA, Woodward L, Adamson ED, Farquhar brane proteins and the cytoplasm. J Biol Chem 267: 8703–8706, MG: The expression of megalin (gp330) and LRP diverges 1992 during F9 cell differentiation. J Cell Sci 108: 1433–1441, 1995 39. Masternak K, Barras E, Zufferey M, Conrad B, Corthals G, 22. Miettinen A, Dekan G, Farquhar MG: Monoclonal antibodies Aebersold R, Sanchez JC, Hockstrasser DF, Mach B, Reith W: A against membrane proteins of the rat glomerulus: Immunochemi- gene encoding a novel RFX-associated trans-activator is mutated cal specificity and immunofluorescence distribution of the anti- in the majority of MHC class II deficiency patients. Nat Genet gens. Am J Pathol 137: 929–944, 1990 20: 273–277, 1998 23. Shi PY, Maizels N, Weiner AM: Recovery of soluble, active 40. Nagarajan UM, Louis-Plence P, DeSandro A, Nilsen R, Bushey recombinant protein from inclusion bodies. Biotechniques 23: A, Boss JM: RFX-B is the gene responsible for the most common 1036–1038, 1997 cause of the bare lymphocyte syndrome, an MHC class II im- 24. Orlando RA, Farquhar MG: Functional domains of the receptor- munodeficiency. Immunity 10: 153–162, 1999 associated protein (RAP). Proc Natl Acad Sci USA 91: 3161– 41. Lin JH, Makris A, McMahon C, Bear SE, Patriotis C, Prasad VR, 3165, 1994 Brent R, Golemis EA, Tsichlis PN: The ankyrin repeat-contain- 25. Orlando RA, Farquhar MG: Identification of a cell line that ing adaptor protein Tvl-1 is a novel substrate and regulator of expresses a cell surface and a soluble form of the gp330/receptor- Raf-1 kinase. J Biol Chem 274: 14706–14715, 1999 associated protein (RAP) Heymann nephritis antigenic complex. 42. Trommsdorff M, Borg JP, Margolis B, Herz J: Interaction of Proc Natl Acad Sci USA 90: 4082–4086, 1993 cytosolic adaptor proteins with neuronal apolipoprotein E recep- 26. Lundstrom M, Orlando RA, Saedi MS, Woodward L, Kurihara tors and the amyloid precursor protein. J Biol Chem 273: 33556– H, Farquhar MG: Immunocytochemical and biochemical char- 33560, 1998 acterization of the Heymann nephritis antigenic complex in rat 43. Kerjaschki D, Farquhar MG: Immunocytochemical localization L2 yolk sac cells. Am J Pathol 143: 1423–1435, 1993 of the Heymann nephritis antigen (GP330) in glomerular epithe- 27. De Vries L, Mousli M, Wurmser A, Farquhar MG: GAIP, a lial cells of normal Lewis rats. J Exp Med 157: 667–686, 1983 protein that specifically interacts with the trimeric G protein G 44. Sedgwick SG, Smerdon SJ: The ankyrin repeat: A diversity of alpha i3, is a member of a protein family with a highly interactions on a common structural framework. Trends Biochem conserved core domain. Proc Natl Acad Sci USA 92: 11916– Sci 24: 311–316, 1999 11920, 1995 45. Nelson WJ, Veshnock PJ: Ankyrin binding to (Na2ϩ ϩ 28. De Vries L, Lou X, Zhao G, Zheng B, Farquhar MG: GIPC, a Kϩ)ATPase and implications for the organization of membrane PDZ domain containing protein, interacts specifically with the C domains in polarized cells. Nature 328: 533–536, 1987 terminus of RGS-GAIP. Proc Natl Acad Sci USA 95: 12340– 46. Rader K, Boyer AD, Farquhar MG, Arden KC: Assignment(1) of 12345, 1998 ankyrin repeat family A (RFXANK-like) 2 (ANKRA2) to human 29. De Vries L, Elenko E, McCaffery JM, Fischer T, Hubler L, Mc- chromosome 5q12–Ͼq13 by radiation hybrid mapping and so- Quistan T, Watson N, Farquhar M: RGS-GAIP, a GTPase-activat- matic cell hybrid PCR. Cytogenet Cell Genet 89: 164–165, 2000 2178 Journal of the American Society of Nephrology J Am Soc Nephrol 11: 2167–2178, 2000

47. Rogers S, Wells R, Rechsteiner M: Amino acid sequences com- 49. Michaely P, Kamal A, Anderson RG, Bennett V: A requirement mon to rapidly degraded proteins: The PEST hypothesis. Science for ankyrin binding to clathrin during coated pit budding. J Biol 234: 364–368, 1986 Chem 274: 35908–35913, 1999 48. Givan SA, Sprague GF Jr: The ankyrin repeat-containing protein 50. Oldenhof J, Vickery R, Anafi M, Oak J, Ray A, Schoots O, Pawson Akr1p is required for the endocytosis of yeast pheromone recep- T, von Zastrow M, Van Tol HH: SH3 binding domains in the tors. Mol Biol Cell 8: 1317–1327, 1997 dopamine D4 receptor. Biochemistry 37: 15726–15736, 1998