Cubilin, a 456 kDa multipurpose receptor lacking in both transmembrane and cytoplasmic domains is expressed in the apical BBMs (brush border membranes) of polarized epithelia. Cubilin interacts with two transmembrane proteins, AMN, a 45–50 kDa protein product of the amnionless gene, and megalin, a 600 kDa giant endocytic receptor. In vitro, three fragments of cubilin, the 113-residue N-terminus and CUB domains 12–17 and 22–27, demonstrated Ca2+-dependent binding to megalin. Immunoprecipitation and immunoblotting studies using detergent extracts of rat kidney BBMs revealed that cubilin interacts with both megalin and AMN. Ligand (intrinsic factor–cobalamin)-affinity chromatography showed that in renal BBMs, functional cubilin exists as a complex with both AMN and megalin. Cubilin and AMN levels were reduced by 80% and 55–60% respectively in total membranes and BBMs obtained from kidney of megalin antibody-producing rabbits. Immunohistochemical analysis and turnover studies for cubilin in megalin or AMN gene-silenced opossum kidney cells showed a significant reduction (85–90%) in cubilin staining and a 2-fold decrease in its half-life. Taken together, these results indicate that three distinct regions of cubilin bind to megalin and its interactions with both megalin and AMN are essential for its intracellular stability.
- amnionless gene
- brush border membrane
- gene silencing
Cubilin is a 456 kDa multidomain  multipurpose receptor  expressed in the apical BBMs (brush border membranes) of several tissue-derived polarized epithelial cells including ileum, renal proximal convoluted tubules, visceral yolk sac and placenta [3,4]. Its well-characterized ligands include IF–Cbl (intrinsic factor–cobalamin) complex, albumin, transferrin and apolipoprotein A1 [5,6]. It contains three distinct regions, a 113-residue N-terminus, eight EGF (epidermal growth factor)-like domains and a continuous stretch of 27 CUB domains. Cubilin however lacks a transmembrane and cytoplasmic domain  and thus internalization of albumin in the renal proximal tubular cells [7,8] and apolipoprotein A1 in the bile duct epithelial cells respectively  is thought to be mediated via its interaction with megalin, a large endocytic receptor of 600 kDa [10–12]. In support of this observation are in vitro studies  that have shown high-affinity binding of cubilin to megalin and the ability of this complex to also bind one of its ligands, IF–Cbl. Furthermore, a number of studies have also shown, using ultrastructural imaging techniques, that cubilin and megalin are co-localized within the apical endosomes and apical BBMs of epithelial cells [1,7,9,13].
In contrast with its interactions with megalin, cubilin interactions with the product of the amnionless gene, AMN, a 45–50 kDa cysteine-rich transmembrane protein, has recently been shown to be essential for the cell-surface expression of cubilin fragments expressed in vitro in cultured cells [14,15]. In support of this observation are results from patients who develop megaloblastic anaemia due to poor absorption of Cbl caused by mutations in the AMN but not cubilin molecule [16,17]. Cbl deficiency due to poor absorption is also noted in patients with mutations in the cubilin molecule itself [18,19].
These latter studies have clearly demonstrated that interactions of AMN with cubilin are important for its anterograde trafficking. However, it is not clear how AMN assists in the apical delivery of cubilin. In addition, nothing is known as to what role, if any, megalin plays in the membrane trafficking of cubilin. The current studies using renal proximal tubular OK (opossum kidney) cells and kidney tissues from rats and rabbits were undertaken to address some of these issues. The results of the present study show that the stability of full length mature cubilin in OK cells is determined by its interactions with both megalin and AMN and that in the native kidney BBM, cubilin exists as a complex with both megalin and AMN.
The following were commercially purchased as indicated: TRIzol® reagent, CNBr (cyanogen bromide)-activated Sepharose and Sepharose–Protein A (Sigma–Aldrich); M-MLV (moloney-murine-leukaemia virus) reverse transcriptase and TNT-T7 quick-coupled transcription translation system (Promega); Ultrafree-MC spin filters (Millipore); pSec Tag B expression vector, Lipofectamine™ 2000 reagent, Opti-MEM reduced-serum medium for siRNA (small interfering RNA) transfection, DMEM (Dulbecco's modified Eagle's medium), methionine-free DMEM and FBS (fetal bovine serum) (Invitrogen); Optitran-BA-reinforced nitrocellulose transfer membranes (MIDSci); Supersignal West Femto maximum sensitivity substrate (Pierce Biotechnology); Lab-Tek Chamber slides (21 mm×45 mm) for immunohistochemistry (Nalge Nunc); [35S]Express protein labelling mix (1175 Ci/mM; PerkinElmer Life Sciences); [57Co]Cbl (10.6 Ci/6 ml (MP Biomedicals LLC); and FluoroHance™ (Research Products).
Megalin was purified to homogeneity from rat kidney as described previously . Antisera to purified rat renal cubilin  and megalin  and rat intestinal alkaline phosphatase  were raised in rabbits as described. Antiserum to human AMN extracellular domain peptide (amino acid residues R191FHGPGALSVGPE DCADPS209) was commercially obtained from Washington Biotech. The antibody was confirmed to be AMN specific as it produced a single band of ∼45 kDa on immunoblotting of various rat or human tissues and a doublet of 45/50 kDa in OK cell membranes.
OK cells were routinely cultured in DMEM (25 mM glucose) supplemented with 20% (v/v) heat-inactivated FBS, 1 mM sodium pyruvate, 2 mM glutamine, 0.25 mM non-essential amino acids and antibiotics (100 units/ml penicillin and 100 μg/ml streptomycin) in a humidified atmosphere containing 5% CO2 at 37 °C. For siRNA transfection experiments, OK cells were cultured in DMEM supplemented with 20% (v/v) FBS without antibiotics and pulse–chase experiments were conducted in methionine-free medium.
Construction of plasmids of rat cubilin fragments
Total RNA from rat kidney was isolated using TRIzol® reagent according to the manufacturer's instructions. Cubilin cDNA fragments covering the entire molecule were generated by RT (reverse transcriptase)–PCR using primer sets described previously . The PCR-amplified products were gel-purified using spin filters and cloned in expression vector pSec Tag B. The authenticity of the amplified cubilin fragments and the direction of the cloned insert were confirmed by sequencing. The plasmids were cell-free translated in vitro by the TNT-T7 quick-coupled transcription and translation system according to the manufacturer's instructions. The [35S]-translated products were used for binding to the affinity matrix containing bound rat megalin.
Binding of cubilin fragments to megalin
Purified megalin was coupled to CNBr-activated Sepharose beads according to the manufacturer's instructions. The [35S]-labelled cell-free translated cubilin fragments (∼50000 dpm) were incubated with 50 μl of a 1:1 suspension of Sepharose–megalin previously equilibrated with buffer A [10 mM Tris/HCl (pH 7.4), 140 mM NaCl, 0.1 mM PMSF, 20 mM benzamidine] containing 10 mM CaCl2 or 10 mM EDTA for 60 min at 25 °C. Following incubation, Sepharose beads were washed exhaustively with buffer A and the bound radioactive products were released by boiling the beads with SDS sample buffer and were subjected to non-reducing SDS/PAGE (12% gels). Gels were fixed for 60 min, treated with FluoroHance™ for 30 min (according to the manufacturer's instructions), and the bands were visualized by fluorography.
Cubilin interaction with AMN and megalin in renal total and apical BBMs
Renal BBMs were prepared from normal rats and from normal and megalin-antibody-producing rabbits using the Ca2+ aggregation method as described previously . Total tissue membranes that include both plasma and intra-organelle membranes were prepared as follows. The tissues were cut into small pieces and were homogenized in buffer A with a loose-fitting Potter–Elvejhem homogenizer with 10 strokes up and down. The homogenate was centrifuged for 30 min at 100000 g and the particulate fraction was resuspended in buffer A. Triton X-100 extracts from rat renal BBMs (50 μg of protein) were incubated for 2 h at 22 °C with 10 μl of cubilin, megalin or AMN antibodies and immunoprecipitated by adding 50 μl of a 1:1 suspension of Sepharose–Protein A in buffer A. After 2 h incubation, the Sepharose beads were washed exhaustively with buffer A and the bound proteins were released by boiling with SDS sample buffer and were subjected to non-reducing SDS/PAGE (5% gels). The separated proteins were transferred for 18 h at 4 °C at a constant voltage of 30 V. The nitrocellulose membranes were then probed for 60 min with diluted (1:1000) antisera to cubilin or megalin followed by incubation for 60 min with a goat anti-rabbit IgG labelled with horseradish peroxidase (1:30000 dilution). Immunoreactive bands were visualized by enhanced chemiluminescence.
Total membranes and apical BBMs (25 μg) from normal and megalin-antibody-producing rabbit kidneys were subjected to SDS/PAGE and immunoblotting for cubilin, megalin, AMN and alkaline phosphatase. For immunoblotting AMN and alkaline phosphatase, BBM proteins were separated on a 7.5% non-reducing gel and transferred on to nitrocellulose membranes for 1 h at 4 °C at a constant voltage (90 V). The rest of the immunoblotting procedure was similar to that described for cubilin or megalin except that the alkaline phosphatase antibody was used at a 1:500 dilution. To compare the protein profiles, apical BBM proteins (50 μg) from normal and megalin-antibody-producing rabbits were separated on a 4–15% gradient gel and stained with Coomassie Brilliant Blue dye. The EDTA-inhibitable binding of rat IF–[57Co]Cbl by total membrane homogenates was determined as described earlier . Briefly, total membrane homogenates from normal and megalin-antibody-producing rabbits were incubated in a volume of 1 ml containing buffer A supplemented with 10 mM CaCl2 or EDTA and rat IF–[57Co]Cbl (2.0 pM). After incubation for 1 h at 22 °C, the bound and free IF–[57Co]Cbl was separated by centrifugation and counted on a γ-counter.
Rat gastric extract containing gastric IF was prepared as described previously , and a 20% (w/v) homogenate was routinely observed to have a Cbl-binding capacity of ∼18–20 ng/ml. Gastric extract (10 ml) was incubated overnight at 4 °C with 1 ml of Sepharose beads covalently linked to Cbl. IF–Cbl–Sepharose beads (ligand-affinity matrix) were exhaustively washed with buffer A and suspended in the same buffer. Triton X-100 extracts from rat renal apical BBMs (10 and 20 μg of protein) were incubated with ligand-affinity matrix (60 μl of a 1:1 suspension in buffer A containing 10 mM CaCl2) for 18 h at 4 °C. Beads were then collected by centrifugation (6000 g for 10 min at 4 °C) and washed with 15–20 ml of buffer A containing 10 mM CaCl2, and the bound proteins were released by boiling with SDS sample buffer and subjected to non-reducing SDS/PAGE. Proteins bound to ligand-affinity matrix or the unbound proteins present in the Triton X-100 extracts (10 and 20 μg of protein) similarly separated on SDS/PAGE were then immunoblotted for cubilin, megalin or AMN. The protein bands visualized on immunoblots were quantified using the Kodak Molecular imaging software on a film-free Kodak Image Station (2000 MMT). At each concentration of the protein used, the percentage recovery of megalin, cubilin and AMN from the ligand-affinity matrix was determined and the ratio of megalin/cubilin or AMN/cubilin recovered was calculated.
Silencing of AMN and megalin gene expression in OK cells
siRNA sequences of the type (AA-N19) were selected from the 5′ region of human cubilin or AMN genes using the on-line Invitrogen siRNA design tool and were modified by the addition of dTdT overhangs at the 3′ ends to increase their stability. The sequences used were Megalin-siRNA1, 5′-AAGAGGCATCAGCAGCGTAAT[dT][dT]-3′; Megalin-siRNA2, 5′-AAGCGGAAGAACCTGTGGTAT[dT][dT]-3′; AMN-siRNA1, 5′-AAGGTCACGCCGTCTCAGACA[dT][dT]-3′; and AMN-siRNA2, 5′-AACCTGCCGTCTTCCGCGACT[dT][dT]-3′. A scrambled siRNA, 5′-AACGCGCCCAGAGCGCAG CTC[dT][dT]-3′, was designed to be used as a negative control.
Transient transfection of megalin and AMN siRNA in OK cells
OK cells were cultured in 60 mm culture dishes containing DMEM supplemented with 20% (v/v) FBS for 24 h to achieve 60–70% confluence, and were transfected with megalin or AMN siRNA using Lipofectamine™ 2000 according to the manufacturer's instructions. Briefly, 100 and 200 pM of siRNAs and 5 μl of Lipofectamine™ 2000 reagent were separately diluted in 250 μl of Opti-MEM reduced-serum medium. After 5 min of incubation at 22 °C, the diluted reagents were mixed and incubated further for 20 min at 22 °C. Cultured cells containing 2.5 ml of freshly added DMEM were treated with siRNA–Lipofectamine™ complex and incubated at 37 °C for 48 h. In another set of dishes, OK cells were transfected with scrambled siRNA or Lipofectamine™ 2000 alone. Total RNA was extracted using TRIzol® reagent and semi-quantitative RT–PCR amplification of AMN and megalin gene fragments was performed using primers AMN Fw, 5′-GGCGAGATGGGCGTCCTGGGC-3′; AMNRev, 5′-GGCCTCGGCCTC GGCCCCGGC-3′; MEGFw, 5′-GAGATGGATCGCGGGCCGGCAGCA-3′; and MEGRev, 5′-AGCCTGGTAGTACACTTGTGG-3′. The specificity of gene silencing was confirmed by RT–PCR amplification of GAPDH (glyceraldehyde-3-phosphate dehydrogenase) using primers GAPFw, 5′-ATCATCCCTGCCTCTA CTGG-3′ and GAPRev, 5′-TGGGTGTCGCTGTTGAAGTC-3′. In another set of dishes, OK cells were harvested and homogenized in buffer A. Equal amounts of total membrane protein (20 μg) were subjected to SDS/PAGE and immunoblotting for AMN and megalin.
Our initial experiments showed that 200 pM of Megalin-siRNA1 and AMN-siRNA2 when transfected for 48 h were found to be most efficient in silencing the respective genes and thus were used for further experiments.
Immunohistochemical staining for cubilin
OK cells were cultured in chamber slides containing 2 ml of DMEM supplemented with 20% (v/v) FBS for 24 h to achieve 60–70% confluence and transfected with megalin and AMN siRNA as described above. The surface area per well of the chamber slides (9.45 cm2) was roughly half the surface area of 60 mm culture dishes (20 cm2), hence the volumes of the transfection reagent, growth medium and siRNA were reduced proportionately. After 48 h incubation, the medium was discarded and cells were washed twice with ice-cold PBS. The chambers were carefully peeled off and the slides were immersed in acetone maintained at −20 °C for 5 min and air-dried. Chamber slide sections were immunostained using the ES autostainer (Ventana Medical Systems) according to the manufacturer's instructions. Sections were antigen-retrieved using citric acid (pH 6.0) in a microwave oven followed by incubation with rat cubilin antiserum at a 1:50 dilution for 32 min. Antigen–antibody complexes were detected with an IVIEW-DAB (diaminobenzidine) detection system (Ventana Medical Systems). The sections were counterstained with haematoxylin and examined by light microscopy on a Leica DMI 4000 B microscope equipped with a Retiga Exi IEEE 1394 FireWire(tm) Digital CCD colour cooled camera (Q imaging). The signal intensity for the cubilin receptor (represented by the yellow colour overlay) was quantified using Image-Pro 6.2 software (Media Cybernetics) and expressed as average counts per two random fields.
Half-life of cubilin
Pulse-chase labelling of OK cells was carried out to determine the half-life of cubilin in AMN or megalin-silenced cells essentially as described previously . Briefly, OK cells were incubated in 3 ml of methionine-free medium for 30 min at 37 °C. The medium was removed and cells were pulse-labelled for 1 h by adding a fresh 3 ml of methionine-free medium containing 100 μCi/ml of [35S]methionine. Labelled cells were then chased with non-radioactive methionine (10 mM) for 0–48 h. At each time interval, cells were harvested and Triton X-100 extracts of the total cell membranes from untransfected, AMN- and megalin-silenced cells were subjected to immunoprecipitation for cubilin using a monospecific cubilin antibody as described above. Immune pellets were exhaustively washed with buffer A and the radioactivity associated with pellets was released by boiling with SDS sample buffer and subjecting to non-reducing SDS/PAGE (5% gels) and fluorography. An aliquot of released radioactivity at each time interval was counted in a γ-counter to determine the half-life of cubilin decay.
Purity of megalin and the specificity of its antiserum
Since some experiments in the present study have used megalin purified from rat kidney as well as its antiserum raised in rabbits, it was important to initially establish the purity of the megalin preparation and the specificity of its antiserum. Protein staining of the purified megalin revealed a single band with a molecular mass of 600 kDa (Figure 1, lane 1) and immunoblotting of rat kidney apical BBMs using antiserum raised to this protein recognized a single immunoreactive band with a molecular mass of 600 kDa (Figure 1, lane 2). These results clearly established the purity of megalin and the specificity of its antiserum.
Three regions of cubilin demonstrate Ca2+-dependent binding to megalin
Although full length cubilin protein is known to bind to megalin with high affinity , little is known about the regions of cubilin that are involved in binding. This issue has been addressed by generating cDNA fragments covering all of the structural domains of cubilin (Figure 2A). SDS/PAGE analysis of the [35S]-labelled cubilin fragments that bound to the Sepharose–megalin affinity matrix revealed that fragments containing the 113-residue N-terminus, eight EGF-like domains and CUB domains 1 and 2, CUB domains 12–17 and CUB domains 22–27, bound to megalin (Figure 2B). Binding by all of the three cubilin fragments to megalin was Ca2+-dependent (Figure 2B, lanes 1, 7 and 11) as, in the presence of EDTA, binding was considerably reduced (Figure 2B, lanes 2, 8 and 12). Specificity of binding to megalin is borne out by the observations that cubilin fragments containing CUB domains 3–8, 8–12 or 17–22 did not show binding. Within the 113-residue N-terminus fragment (Figure 2C, lanes 3 and 4), a 64 residue fragment derived from residues 50–113 (Figure 2C, lanes 1 and 2) demonstrated binding to megalin, but the binding was not Ca2+ dependent. However, Ca2+-dependency of megalin binding by the 113-residue N-terminus required the presence of both CUB domains 1 and 2 and EGF-like repeats (Figure 2C, lanes 7 and 8), although the eight EGF-like repeats by themselves did not bind megalin (Figure 2C, lanes 5 and 6).
Cubilin co-immunoprecipitates with AMN and megalin when treated with antisera to megalin and AMN
In the present and previous  studies, interactions between cubilin and megalin have been shown to occur in vitro where both proteins are present in a membrane-free form. However, in the renal and intestinal BBMs, cubilin is not free and is bound to AMN as a complex known as CUBAM . Thus, in the native BBMs, interactions of megalin must occur with the CUBAM complex. In order to explore this possibility, immunoblotting for cubilin and megalin was carried out using immunoprecipitated material obtained from Triton X-100 extracts of renal BBMs using antiserum raised to megalin or AMN and cubilin or AMN respectively. The results show that antisera to both megalin and AMN could immunoprecipitate cubilin (Figure 3, left-hand panels) and that antisera to both cubilin and AMN could immunoprecipitate megalin (Figure 3, right-hand panels). The specificity of these pulldown assays is evident as none of the antisera used for immunoprecipitation could immunoprecipitate alkaline phosphatase, a component of rat renal BBMs (results not shown).
Megalin interacts with CUBAM to form a complex and can be purified by ligand-affinity chromatography
Megalin, along with cubilin and AMN could be detected as a single band by immunoblotting 10 μg (Figure 4, upper panels, lanes 1) and 20 μg (Figure 4, upper panels, lanes 2) of free Triton X-100 extract of rat renal BBM or that bound to Sepharose–IF–Cbl affinity matrix (Figure 4, upper panels, lanes 3 and 4). Quantification of immunoreactive bands revealed that the percentage recovery from the ligand-affinity matrix (Figure 4, bottom panels) varied from 80–95% for cubilin, megalin or AMN and the ratios of percentage recovery between megalin and cubilin or AMN and cubilin were 0.85–0.95. Taken together, these ligand-binding studies showed that within the native BBMs, functional cubilin is a complex containing both AMN and megalin. The affinity matrix by itself, when subjected to SDS/PAGE and stained for protein, revealed a single protein with a molecular mass of 48 kDa and this band immunoreacted to an antibody raised against rat gastric IF (results not shown). These results clearly indicated that the CUBAM–megalin complex obtained from the affinity matrix was bound to IF, a ligand for cubilin, but not megalin .
Cubilin and AMN protein levels are significantly reduced in the renal membranes of rabbits producing megalin antiserum
Previously  we have shown that megalin exists as an immune complex in the renal BBMs of rabbits producing high titre antibodies to megalin. Although megalin protein levels were unaltered in normal and megalin-antibody-producing rabbits, there was a significant increase in the urinary excretion of many megalin ligands . Immunoblotting for cubilin using kidney total membranes (Figure 5A) or BBMs (Figure 5D), or AMN using total membranes (Figure 5F) or BBMs (Figure 5G) revealed significant decreases in the levels of both of these proteins. Cubilin activity (Figure 5B) and protein levels (Figure 5C) in the total membranes decreased by 80% and similar levels of decrease in both were also noted using apical BBMs (results not shown). AMN protein levels decreased by approx. 55–60% in both the total membranes (Figure 5E) and BBMs (Figure 5H). The decrease in cubilin and AMN levels in megalin-antibody-producing rabbits is specific to megalin-related proteins. Overall, the protein profile (Figure 5I) in the apical BBMs of control (Figure 5I, lane 1) and megalin-antibody-producing rabbit (Figure 5I, lane 2) kidney remained largely unaltered. In addition, immunoblotting for alkaline phosphatase (Figure 5J), a protein unrelated to megalin, showed no changes in the BBMs of control (Figure 5J, lane 1) and megalin-antibody-producing (Figure 5J, lane 2) rabbits. Taken together, these results indicated that the blocking of the protein-binding sites of megalin by its antibody seems to decrease only the levels of cubilin and AMN and that the decreases in cubilin levels may be due to increased intracellular degradation. In order to test the effect of the lack of megalin or AMN binding on the stability of cubilin, we determined cellular staining and changes in the intracellular turnover of cubilin in renal OK cells that were silenced for AMN or megalin genes.
Effect of AMN and megalin gene silencing on cellular staining and intracellular turnover of cubilin
Gene-silencing studies and semi-quantitative RT–PCR (Figure 6A) revealed that using 100 (Figure 6A, lane 3) or 200 (Figure 6A, lane 4) pM of specific siRNA for megalin (Figure 6A, top panel) and AMN (Figure 6A, middle panel), transcript levels of both declined by approx. 75–85% relative to their levels in untransfected (Figure 6A, lane 1) or scrambled siRNA-transfected cells (Figure 6A, lane 2). In these cells, the mRNA levels of GAPDH, an internal control, did not change (Figure 6A, bottom panel). Immunoblotting (Figure 6B) using total membranes from untransfected (Figure 6B, lane 1) or scrambled siRNA-transfected (Figure 6B, lane 2) or 100 (Figure 6B, lane 3) or 200 (Figure 6B, lane 4) pM of specific siRNA for megalin and AMN revealed loss of megalin (Figure 6B, top panel) and AMN (Figure 6B, bottom panel) proteins confirming the silencing of specific genes in these cells. Further confirmation for the loss of cubilin in these cells was obtained by immunohistochemical analysis using a polyclonal antiserum to rat renal cubilin (Figure 6C). Relative to untransfected and scrambled siRNA-transfected cells, the signal intensities of cubilin was drastically reduced in both AMN- and megalin-silenced cells. Average counts of signal intensities for cubilin (Figure 6D) from duplicated fields revealed an approx. 90–95% reduction in AMN- (Figure 6D, bar 3) or megalin- (Figure 6D, bar 4) silenced cells relative to untransfected (Figure 6D, bar 1) or siRNA-transfected cells (Figure 6D, lane 2). In megalin-silenced OK cells (Figure 7A), the half-life of cubilin was reduced to 15–18 h and in AMN-silenced OK cells the half-life of cubulin was reduced to 20–22 h, indicating that cubilin degradation is more rapid in megalin- or AMN-silenced cells. SDS/PAGE (Figure 7B) of immunoprecipitated [35S]cubilin revealed that the 456 kDa cubilin levels were drastically reduced within 24 h of chase as compared with [35S]cubilin from untransfected cells for the same length of chase time.
In the present study we have provided evidence from a number of experiments that megalin interacts with cubilin both in vitro and in vivo. There are several important implications from our in vitro studies. First, these studies provide evidence that cubilin binding to megalin occurs at three specific regions of cubilin which involve the 113-residue N-terminus and CUB domains 12–17 and 22–27. Secondly, interactions of the N-terminus of cubilin with megalin appear to be physiologically relevant as its truncation results in destabilization of cubilin in thyroidectomized rats . In addition, the N-terminus could also be involved in the membrane stabilization of cubilin via interactions with the lipid bilayer involving either its amphipathic α-helix or a palmitoyal residue localized to this region. Previously  we have shown that cubilin expressed in OK cells is indeed palmitoylated. Thirdly, the significance of megalin binding by CUB domains 12–17 or 22–27 is not known. Megalin binding by CUB domains 12–17 may be regulated in vivo by a receptor-associated protein which in vitro is known to compete for cubilin binding . Megalin binding by CUB domains 22–27 may provide an additional scaffolding effect for stabilization of cubilin. Finally, lack of megalin binding by CUB domains 3–8 may allow binding of various other cubilin ligands such as the IF–Cbl complex, albumin, transferrin and vitamin D-binding protein [5,6], as cubilin binding to these ligands is localized in CUB domains 3–8.
Although these in vitro studies have shown direct binding of certain specific regions of cubilin to megalin, it is possible that in the native membranes, the interaction may be mediated via a common ligand that binds to both cubilin and megalin. However, a cubilin–megalin interaction via a common ligand is unlikely since the results of the present study show (Figure 4) that megalin is eluted quantitatively from an affinity matrix linked to IF, a ligand only for cubilin, but not megalin . In addition to megalin, previous studies [14,15] have shown that cubilin exists as a complex with AMN and in megalin-deficient cells transfected with an N-terminal fragment of cubilin, AMN assisted in its cell-surface expression. Interestingly, the cubilin fragment was also able to mediate endocytosis of the ligand IF–Cbl. These studies have indicated that the anterograde and retrograde trafficking of a cubilin fragment that included the N-terminus may be independent of megalin. It is remarkable that this cubilin fragment is able to achieve both transport and functional competence within cells. It is probable that cubilin is able to do so since overall folding into its mature full length form is hierarchal, proceeding from secondary structure via subdomains and domains towards attainment of its complete tertiary structure. However, it is not clear how AMN is able to mediate the bidirectional membrane trafficking of the cubilin fragment and whether this is the only role of AMN in a somatic cell. This latter consideration is important since AMN is expressed at sites where cubilin is not expressed and cubilin is expressed at sites where AMN is not expressed [9,17,26–29]. These observations suggest that AMN, under certain circumstances, can take over and mediate both anterograde and retrograde trafficking of cubilin. However, it is not clear whether AMN can function alone in polarized epithelia known to express high levels of megalin.
Our present study demonstrating the existence of a cubilin–AMN–megalin complex in the renal BBMs suggests that megalin binds to CUBAM and the stoichiometry of binding is approx. 1:1. However, it is important to note that quantification of data based on relative intensities of immunoreactive bands is at best approximate. Our studies using megalin-antibody-producing rabbits show significant reductions in AMN and cubilin protein levels in both the apical BBMs as well as in the total membranes which is a mixture of plasma and intraorganelle membranes. Thus it is likely that the lack of interaction with megalin results in the degradation of the CUBAM complex from within the cells which results in its deficit in the apical BBMs. This suggestion is supported by our turnover data which show increased turnover of cubilin in megalin-silenced cells. However, other sites of CUBAM degradation, due to its inability to bind megalin such as plasma membrane or during endocytosis and recycling, cannot be ruled out.
Previously  we have shown in the renal BBMs of these rabbits that megalin exists as an immune complex and that urinary excretion of many megalin ligands such as albumin and transcobalamin-II is increased. The mechanism causing significant reduction in cubilin and AMN protein levels in megalin-antibody-producing rabbits or of cubilin staining in megalin- or AMN-silenced cells is not known. In addition, due to lower magnification of immunohistochemical detection of cubilin used in the present study, the possibility that cubilin may be retained intracellularly in AMN- or megalin-silenced cells cannot be ruled out. Despite these caveats, our turnover studies do suggest that cubilin may be degraded faster in cells silenced for the AMN or megalin gene. Thus mutual interaction between AMN, cubilin and megalin may be important in the intracellular stability of the CUBAM complex. This may help explain why in some megaloblastic anaemia patients there is a lack of cell-surface expression of cubilin due to mutations in the AMN molecule [16,17,30]. It is possible that lack of cubilin binding to AMN due to a variety of mutations in AMN could result in its misfolding which in turn could affect its binding to megalin resulting ultimately in its enhanced degradation. Alternatively, it is possible that lack of megalin expression may disrupt the entire endocytic apparatus and the internalized CUBAM complex may be diverted to the degradative pathway. In support of this suggestion are results from megalin knockout mice that have shown disruption of endocytic compartments [31,32] and involvement of megalin in the endocytosis of various cubilin ligands [33,34].
In summary, the results of the present study show that megalin interacts in vitro and in vivo with cubilin and the CUBAM complex respectively, and that these interactions appear to be essential for the intracellular stability as well as the surface expression of cubilin. Future work should focus on further defining the structural basis for these interactions in the anterograde and retrograde trafficking of cubilin.
This work was supported by a grant from the Department of Veteran Affairs (7816-01P) awarded to B. S.
Abbreviations: AMN, amnionless; BBM, brush border membrane; Cbl, cobalamin; CUBAM, cubulin–AMN complex; DMEM, Dulbecco's modified Eagle's medium; EGF, epidermal growth factor; FBS, foetal bovine serum; GAPDH, glyceraldehyde-3-phosphate dehydrogenase; IF, intrinsic factor; OK, opossum kidney; RT, reverse transcriptase; siRNA, small interfering RNA
- © The Authors Journal compilation © 2008 Biochemical Society