Basolateral sorting of the coxsackie and adenovirus receptor through interaction of a canonical YXXPhi motif with the clathrin adaptors AP-1A and AP-1B

Abstract

The coxsackie and adenovirus receptor (CAR) plays key roles in epithelial barrier function at the tight junction, a localization guided in part by a tyrosine-based basolateral sorting signal, (318)YNQV(321). Sorting motifs of this type are known to route surface receptors into clathrin-mediated endocytosis through interaction with the medium subunit (μ2) of the clathrin adaptor AP-2, but how they guide new and recycling membrane proteins basolaterally is unknown. Here, we show that YNQV functions as a canonical YxxΦ motif, with both Y318 and V321 required for the correct basolateral localization and biosynthetic sorting of CAR, and for interaction with a highly conserved pocket in the medium subunits (μ1A and μ1B) of the clathrin adaptors AP-1A and AP-1B. Knock-down experiments demonstrate that AP-1A plays a role in the biosynthetic sorting of CAR, complementary to the role of AP-1B in basolateral recycling of this receptor. Our study illustrates how two clathrin adaptors direct basolateral trafficking of a plasma membrane protein through interaction with a canonical YxxΦ motif.

Fig. 1.

Mutation in CAR's YXXΦ motif alters its basolateral distribution in polarized MDCK cells. (A) Schematic representation of CAR indicating the position and amino acid composition of its YXXΦ motif. (B) Confocal images of MDCK cell lines expressing CAR-GFP WT or V321A, or Y318A (green). (Scale bar, 10 μm.) (C) Domain-selective biotinylation showed more than 90% of WT and V321A CAR at the basolateral PM, in contrast with only 69.9 ± 0.4% and 48.35 ± 6.6% of V321E and Y318A, respectively. The basolateral polarity of Na/K ATPase was conserved in the different cell lines. Histogram bars represent mean ± SEM. *P < 0.05, ***P < 0.001. (D) The extracellular and transmembrane domains of p75 were cloned in frame with CAR's cytoplasmic tail, either WT or carrying the mutations Y318A or V321A. Surface immunofluorescence using a monoclonal antibody against the ectodomain of p75 (green) shows basolateral localization of p75-CAR and apical localization of p75-CARY318A and p75-CARV321A. Nuclei were stained with DAPI (blue) and tight junctions with ZO-1 (red). (Scale bar, 10 μm.) See Table S1 for statistical analysis of all comparisons.

Fig. 2.

Polarized biosynthetic delivery of CAR is driven by its basolateral sorting signal YNQV. (A) Workflow for SBAS quantitative assay. This method allows measurement of the fraction of CAR-GFP delivered to apical or basolateral membranes (CAR avidin-shifted) as a fraction of the total protein synthesized (CAR avidin+biotin unshifted) radiolabeled with a pulse of S³⁵cysteine/methionine. Amount delivered to each surface is calculated as: [intensity of the CAR-GFP band (75 kDa) in minus-Biotin lane] − [intensity of the CAR-GFP band (75 kDa) in plus-Biotin lanes]/(the intensity of the CAR-GFP band in minus-Biotin lane) × 100. This approach is superior to measuring the CAR-GFP displaced to MWs > 75 kDa by streptavidin because these complexes are more difficult to quantify. (B) Mutations V321E and Y318A in CAR's basolateral signal cause intracellular retention of the protein after 120 min relative to WT CAR (white bars; *P < 0.05). (Lower) One representative experiment. Bars represent the mean ± SEM (n ≥ 4). See Table S2 for statistical analysis.

Fig. 3.

CAR interacts with μ1A and μ1B but not with μ3A or μ4 through Y318 and V321. (A) Y2H assays demonstrated direct interactions of CAR's cytoplasmic tail with μ1A and μ1B. Competition with different concentrations of 3-AT shows that the interaction with μ1B was stronger than that with μ1A. (B) Deletion analysis indicates that the interaction of CAR's cytoplasmic tail with μ1A and μ1B involves amino acids 315–344. (C) Y2H assays performed on plates showed that mutations Y318A and V321A but not N319A or Q320A abolish CAR's interaction with μ1A and μ1B. (D) Y2H assays in liquid culture revealed variable interaction strengths between CAR and μ1B depending on the specific V321 substitution. Positive controls of interactions included double transformations with p53 and SV40 T-antigen, and negative controls were obtained by cotransformation of p53 with activation domain constructs and of SV40 T-antigen with BD constructs.

Fig. 4.

CAR's basolateral signal interacts with a conserved tyrosine recognition pocket in μ1A, and μ1B. (A) Sequence alignment analyses demonstrates that most of the amino acids conforming the pocket in AP-2 μ2 that binds YxxΦ motifs are conserved in both μ1A and μ1B. (B) Modeling of μ1A and μ1B based on the known structure of μ2 indicated conservation of the tertiary structure of the pocket, with the exception of K420 in human μ2 which is replaced by P407 in μ1A and μ1B. (C) Mutations F172S and D174S in μ1A and μ1B, involving phenylalanine and aspartic acid residues critical for interaction with Y residues of YxxΦ motifs, block interaction with CAR's cytoplasmic tail. p53 and SV40 T antigen were used as positive and negative controls, as in Fig. 2.

Fig. 5.

AP-1A and AP-1B co-operate in the basolateral sorting of CAR. (A) Steady-state localization. Domain selective biotinylation demonstrates significant loss of basolateral distribution of CAR-GFP in B-KD (68.93 ± 0.65%), even higher in AB-KD (57.15 ± 0.55%) MDCK cells, relative to WT (95.92 ± 1.58%) and A-KD (98.37 ± 1.48%) in MDCK cells. Endogenous Na/K ATPase was significantly depolarized only in AB-KD compared with WT, A-KD, or B-KD MDCK cells. (B) Biosynthetic delivery. After 2 h of biosynthetic delivery, AB-KD cells display 31.60 ± 3.47% of radioactively labeled CAR in the apical membrane, significantly higher than WT (13.34 ± 2.13%), A-KD (16.89 ± 3.17%), and B-KD (11.63 ± 2.11%) MDCK cells. Symbols represent the Mean ± SEM. All points were at least assayed in triplicate. See Tables S3 and S4 for statistical analysis. **P < 0.01, ***P < 0.001.