Do GGA adaptors bind internal DXXLL motifs?

Abstract

The GGA family of clathrin adaptor proteins mediates the intracellular trafficking of transmembrane proteins by interacting with DXXLL-type sorting signals on the latter. These signals were originally identified at the carboxy-termini of the transmembrane cargo proteins. Subsequent studies, however, showed that internal DXXLL sorting motifs occur within the N- or C-terminal cytoplasmic domains of cargo molecules. The GGAs themselves also contain internal DXXLL motifs that serve to auto-regulate GGA function. A recent study challenged the notion that internal DXXLL signals are competent for binding to GGAs. Since the question of whether GGA adaptors interact with internal DXXLL motifs is fundamental to the identification of bona fide GGA cargo, and to an accurate understanding of GGA regulation within cells, we have extended our previous findings. We now present additional evidence confirming that GGAs do interact with internal DXXLL motifs. We also summarize the recent reports from other laboratories documenting internal GGA binding motifs.

Figure 1. Sequence alignment of a number of internal DXXLL motifs

The numbers on the right indicate the C-termini of the proteins. ClC7 is a multi-pass transmembrane protein while consortin is a type-II membrane protein. (h=human, m=mouse, d=drosophila). Residues mutated in this study are shown in bold.

Figure 2. GGA2-DXXLL peptide binding assay

Biolayer interferometry was used to assay the binding of various DXXLL peptides to purified GGA2 (A–D). The association and dissociation of increasing concentrations of the different MBP-peptides to GGA2 are shown (A–D). The affinity (Table 1) of the peptides for GGA2 was calculated after subtraction of the signals obtained with MBP alone. The sequences of the ligands are as follows: CI-MPR – DDSDEDLLHV; LRP9 pLL/dLL – EDEDDVLLL……EAEDEPLLA; pLL/dLL→AA–EDEDDVLLL …….EAEAEPAAA; pLL→AA/dLL→AA–EDEDDVAAA…..EAEAEPAAA

Figure 3. Modeling of complexes between the LRP9 and LRP12 internal DXXLL motifs with the GGA1-VHS domain

Computational docking of the LRP9 and LRP12 internal DXXLL motifs to the VHS domain of human GGA1 or GGA2 (See Materials and Methods for details). As LRP9 contains two overlapping potential DXXLL motifs, the corresponding sequence was docked in two different positions so that LRP9 residues 690 and 689 correspond to the canonical aspartic acid of the motif respectively (panels A and B). Panel (C) shows the docking of the LRP12 DXXLL motif. Panels (D) and (E) show the docking of the LRP9 motif to the GGA2 VHS domain in the same registers as in panels (A) and (B), respectively. In each panel, the upper row shows the conformations of the 10 best models from FlexPepDock (yellow stick representation) docked onto the VHS domain in white (GGA1) or light blue (GGA2) cartoon and surface representations. The rightmost structure in the top row shows the best-scoring peptide model. The bottom row shows closeup views of respectively, 1) the canonical aspartic acid and surrounding residues; and 2) the dileucine motif and additional C-terminal residues. LRP9 and LRP12 residues are shown in stick representation with yellow carbon atoms, GGA1 VHS domain residues are shown with white carbon atoms, and GGA2 VHS domain residues are shown with light blue carbon atoms. Dotted lines indicated putative hydrogen bonds or salt bridges.

Figure 3. Modeling of complexes between the LRP9 and LRP12 internal DXXLL motifs with the GGA1-VHS domain

Computational docking of the LRP9 and LRP12 internal DXXLL motifs to the VHS domain of human GGA1 or GGA2 (See Materials and Methods for details). As LRP9 contains two overlapping potential DXXLL motifs, the corresponding sequence was docked in two different positions so that LRP9 residues 690 and 689 correspond to the canonical aspartic acid of the motif respectively (panels A and B). Panel (C) shows the docking of the LRP12 DXXLL motif. Panels (D) and (E) show the docking of the LRP9 motif to the GGA2 VHS domain in the same registers as in panels (A) and (B), respectively. In each panel, the upper row shows the conformations of the 10 best models from FlexPepDock (yellow stick representation) docked onto the VHS domain in white (GGA1) or light blue (GGA2) cartoon and surface representations. The rightmost structure in the top row shows the best-scoring peptide model. The bottom row shows closeup views of respectively, 1) the canonical aspartic acid and surrounding residues; and 2) the dileucine motif and additional C-terminal residues. LRP9 and LRP12 residues are shown in stick representation with yellow carbon atoms, GGA1 VHS domain residues are shown with white carbon atoms, and GGA2 VHS domain residues are shown with light blue carbon atoms. Dotted lines indicated putative hydrogen bonds or salt bridges.

Figure 4. Residues outside of the core internal DXXLL motif significantly impact binding

(A–C) Pull down assays were performed using either GST or the GST-GGA2 VHS-GAT domain with HEK 293 cell-expressed wild-type (wt) and mutant LRP12 encoding a HA epitope at position 620 within the cytoplasmic tail. 2% of the supernatant (s/n) and 10% of the pellet (p) fractions were loaded for immunoblot analysis (AC) while 2% of the supernatant and 30% of the pellet fractions were loaded for Ponceau Red visualization of the LRP12 band in the affinity pull-down with GGA2 (A-lower panel). Membrane blots were probed with an anti-HA antibody (A–C).

Figure 5. Endogenous GGA2 binds well to the CI-MPR DXXLL signal but not endogenous GGA1 or GGA3

Pull down assays were performed using either GST or the GST-CI-MPR DXXLL peptide with mouse brain cytosol or untransfected HEK 293 cell lysate. A single binding reaction was performed for the control protein (GST) and the ligand peptide (CI-MPR) using the two different lysates as a source of endogenous GGAs. Each pellet fraction was divided into 3 equal parts for SDS-PAGE and Western blot analysis and individually probed for GGA1, GGA2 and GGA3 as described under Materials & Methods. 5% of the GST supernatant was loaded to indicate input.

Figure 6. Mouse GGA1 is autoinhibited similar to its human ortholog

(A–E) Pull down assays were performed using either GST or the GST-CI-MPR DXXLL peptide with HEK 293 cell-expressed wild-type (wt) or mutant mouse GGA1-Flag. All GGA1 proteins are full-length unless otherwise indicated. GGA1 VHS-GAT encodes the first 332 and GGA2 VHS-GAT encodes the first 325 amino acids of their respective full-length proteins. The amino acids mutated within the GGA1 hinge are shown in bold in Figure 1. GST-AP-1 γ-ear was included as a positive control in A. 2% of the supernatant (s/n) and 10% of the pellet (p) fractions were loaded for SDS-PAGE and Western blotting (A–E). The blots were probed with an anti-Flag antibody to detect the wt and mutant GGA1.

Figure 7. The internal DXXLL motif of LRP9 recruits GGA1 to the TGN

Hela cells co-transfected with myc-GGA1, and either wt HA-LRP9 (top two panels) or the C-terminal DXXLL mutant (pLL/dLL→AA) (bottom two panels), were probed with an anti-myc mouse monoclonal antibody or an anti-HA rabbit polyclonal antibody as described in Materials and Methods. Only cells co-expressing GGA1 and LRP9 (wt and mutant) show enhanced Golgi localization of the GGA (arrowheads).