The disabled 1 phosphotyrosine-binding domain binds to the internalization signals of transmembrane glycoproteins and to phospholipids

Abstract

Disabled gene products are important for nervous system development in drosophila and mammals. In mice, the Dab1 protein is thought to function downstream of the extracellular protein Reln during neuronal positioning. The structures of Dab proteins suggest that they mediate protein-protein or protein-membrane docking functions. Here we show that the amino-terminal phosphotyrosine-binding (PTB) domain of Dab1 binds to the transmembrane glycoproteins of the amyloid precursor protein (APP) and low-density lipoprotein receptor families and the cytoplasmic signaling protein Ship. Dab1 associates with the APP cytoplasmic domain in transfected cells and is coexpressed with APP in hippocampal neurons. Screening of a set of altered peptide sequences showed that the sequence GYXNPXY present in APP family members is an optimal binding sequence, with approximately 0.5 microM affinity. Unlike other PTB domains, the Dab1 PTB does not bind to tyrosine-phosphorylated peptide ligands. The PTB domain also binds specifically to phospholipid bilayers containing phosphatidylinositol 4P (PtdIns4P) or PtdIns4,5P2 in a manner that does not interfere with protein binding. We propose that the PTB domain permits Dab1 to bind specifically to transmembrane proteins containing an NPXY internalization signal.

FIG. 1

Association between Dab1 and APP. (A) Retinoic acid treatment of P19 EC cells induces their differentiation into postmitotic neurons and glia and induces increased expression of APP as detected by Western blotting of total cell lysates (lanes 1 through 3; days of retinoic acid treatment are indicated above). Cell lysates were incubated with a GST fusion protein containing the wild-type Dab1 PTB domain (PTB; lanes 4, 6, and 8) or a fusion protein containing the 56E mutant PTB domain (lanes 5, 7, and 9). Bound APP was detected by Western blotting. The values between the gels are molecular sizes in kilodaltons. (B) Vector control (lanes 1 and 2) and wild-type (WT; lanes 3 and 4) and 158V mutant (lanes 5 and 6) Dab1 p80 expressed in 293T cells together with the Myc-tagged APP cytoplasmic domain (mT-APP). Immunoprecipitates were prepared by using preimmune serum (p; lanes 1, 3, and 5) or an anti-Dab1 (B3) antibody (i; lanes 2, 4, and 6) and analyzed by SDS-PAGE and Western blotting with antibodies to Dab1 (upper) or the Myc tag (lower).

FIG. 2

Direct binding between the Dab1 PTB domain and synthetic peptides. Various peptides, 15 to 17 residues in length, were synthesized on cellulose filters in a grided array. Filters were incubated with a ³²P-labelled GSTag-PTB domain fusion protein, and the amount of the bound PTB domain was determined by autoradiography (A) or PhosphorImager (B). (A) Typical filter. (B) Comparison between interacting and noninteracting peptide sequences (left) showing invariant residues (dark shade) and conservative changes (light shade). Percent binding of the PTB domain to the indicated peptides, either unphosphorylated (left histogram) or phosphorylated (right histogram) at tyrosine, indicated by dark shading, is expressed relative to binding to the APP dephosphopeptide (set at 100%). Peptide sequences are designated by protein abbreviations. IR, insulin receptor; EGFR, epidermal growth factor receptor; LDL, low-density lipoprotein receptor.

FIG. 3

Relative binding of the PTB domain to 17-residue peptides based on the APP sequence. (A) Substitution of alanine for each amino acid in turn. The original amino acid is indicated below the histogram. (B) Retesting of each alanine substitution that resulted in reduced binding with 19 amino acids. The positions of the substitutions are indicated in each window by underlining of the wild-type residue. Substituted amino acids are listed below. Phosphotyrosine (pY) was substituted at positions −5 and 0.

FIG. 4

Binding of the GST-PTB domain to the APP C-terminal peptide in solution. (A) Direct binding assay. Increasing amounts of the GST-PTB domain were added to a solution containing a fixed amount of fluorescently labelled APP peptide (probe). The proportion of probe bound to the PTB domain was determined from the fluorescence polarization (squares). The hyperbolic binding curve was fitted with a K_d of 0.55 μM. GST protein alone did not bind (diamond). (B) Nonfluorescent peptide or phosphopeptide competitors added to a mixture of the GST-PTB domain (1.1 μM) and a fixed amount of the fluorescent APP peptide. Addition of increasing amounts of the unphosphorylated APP peptide resulted in a decrease in the fraction of fluorescent APP bound, with half-maximal inhibition at 2 μM (closed squares). The Tyr-0-phosphorylated APP (p.APP) peptide competed less well, with half-maximal inhibition at 500 μM (closed circles). Phosphatase treatment of the phosphopeptide restored its ability to compete with the fluorescent probe for binding (open circle).

FIG. 5

The Dab1 PTB domain interacts with phosphoinositides in a stereospecific manner. (A) GST, GST-Shc PTB domain, and GST-Dab1 PTB domain proteins incubated in the presence (lanes 1 and 2) or absence (lanes 3 and 4) of LUVs containing 66% PtdSer, 17% PtdIns4P, and 17% PtdIns4,5P₂. The LUVs were pelleted by high-speed centrifugation, and the proteins in the supernatant (S) and pellet (P) fractions were detected by Western blotting with anti-GST antibodies. (B) Dab1 PTB domain fusion protein incubated with LUVs (33% each PtdSer, PtdCho, and PtdEth) containing no phosphoinositides, 5% PtdIns4P, 5% PtdIns4,5P₂, or 5% PtdIns3,4,5P₃. Proteins in the supernatant (S) and pellet (P) fractions were detected as described for panel A. The fraction of the GST-PTB domain associated with LUVs was determined in each case. (C) Binding of the Dab1 PTB domain fusion protein to LUVs containing 0 or 5% PtdIns4,5P₂ in the absence (lanes 1 to 4) or presence of various inositol phosphates at 100 μM (d-Ins1,3,4P₃ [lanes 5 and 6], l-Ins1,4,5P₃ [lanes 7 and 8], d-Ins1,4,5P₃ [lanes 9 and 10], d-Ins1,4P₂ [lanes 11 and 12], and d-Ins1P [lanes 13 and 14]). Only d-Ins1,4,5P₃ was an effective competitor of binding. (D) Partial thrombin digestion of the GST-PTB domain fusion released the PTB domain and GST, which were then incubated with LUVs. The liberated PTB domain bound to LUVs containing PtdIns4,5P₂, while the liberated GST did not. Molecular structures are depicted with inositol rings (hexagons), hydroxyl groups (open circles), phosphates (closed circles), and a diacylglycerol backbone (branch). Groups oriented above (large circles) and below (small circles) the plane of the paper are indicated.

FIG. 6

Independence of peptide and phospholipid binding. (A) Reduced peptide binding of the F158V mutant PTB domain. Binding of wild-type (WT; squares) and mutant (circles) GST-PTB domains to a fluorescent APP peptide probe was determined by using fluorescence polarization. The hyperbolic curves were fitted with a K_d of 400 nM (wild type) or 10 μM (F158V mutant). Binding reactions were competed specifically by addition of the nonfluorescent APP peptide (open symbols) but not the phosphorylated APP (p.APP) peptide (shaded symbols). (B) Binding of wild-type (lanes 1 and 2) and F158V mutant (lanes 3 and 4) GST-PTB domains to LUVs containing 5% PtdIns4,5P₂. Assays were performed as described in the legend to Fig. 4. S, supernatant; P, pellet. (C) Binding of fluorescent APP (F-APP) peptide to LUVs mediated by the GST-PTB domain. The fluorescent APP peptide was mixed with LUVs (containing 5% PtdIns4,5P₂) and the wild-type (lanes 1 and 3) or F158V mutant (lane 2) GST-PTB domain in the absence (lanes 1 and 2) or presence (lane 3) of an excess of a nonfluorescent APP peptide competitor. The LUVs were isolated by centrifugation, and the associated fluorescent peptide was quantified.

FIG. 7

Dab1 colocalizes with APP in primary neurons. (A to D) Triple labelling for F-actin (A; blue), Dab1 (B; red), and APP (C; green). D, merged image at higher magnification, showing the axonal growth cone. Overlap of Dab1 and APP appears yellow.