Galectin-1 dimers can scaffold Raf-effectors to increase H-ras nanoclustering

Abstract

Galectin-1 (Gal-1) dimers crosslink carbohydrates on cell surface receptors. Carbohydrate-derived inhibitors have been developed for cancer treatment. Intracellularly, Gal-1 was suggested to interact with the farnesylated C-terminus of Ras thus specifically stabilizing GTP-H-ras nanoscale signalling hubs in the membrane, termed nanoclusters. The latter activity may present an alternative mechanism for how overexpressed Gal-1 stimulates tumourigenesis. Here we revise the current model for the interaction of Gal-1 with H-ras. We show that it indirectly forms a complex with GTP-H-ras via a high-affinity interaction with the Ras binding domain (RBD) of Ras effectors. A computationally generated model of the Gal-1/C-Raf-RBD complex is validated by mutational analysis. Both cellular FRET as well as proximity ligation assay experiments confirm interaction of Gal-1 with Raf proteins in mammalian cells. Consistently, interference with H-rasG12V-effector interactions basically abolishes H-ras nanoclustering. In addition, an intact dimer interface of Gal-1 is required for it to positively regulate H-rasG12V nanoclustering, but negatively K-rasG12V nanoclustering. Our findings suggest stacked dimers of H-ras, Raf and Gal-1 as building blocks of GTP-H-ras-nanocluster at high Gal-1 levels. Based on our results the Gal-1/effector interface represents a potential drug target site in diseases with aberrant Ras signalling.

Figure 1. Galectin-1 neither binds to a farnesylated Ras-peptide nor to H-ras in solution experiments.

(A,B) Fluorescence polarization binding assay of 0.25 μM fluorescein labelled and farnesylated Rheb peptide titrated with increasing concentrations of (A) the farnesyl-binding protein PDEδ or (B) purified Gal-1. (C) Sensitized acceptor FRET binding experiment of 250 nM H-ras and 250 nM Gal-1 or 250 nM C-Raf–RBD (RBD) fluorescently labelled using the ACP-tag technology. The legend to the left shows interaction partners schematically. H-ras was either GTPγS (GTP) or GDP loaded, as indicated. Fluorescent labelling substrates, coenzyme A (CoA)-linked ATTO-488 as a FRET-donor and DY-547 as FRET-acceptor, are represented by green and red stars, respectively. Control sample was a 1:1 mix of fluorescent labelling substrates each at 100 nM. Error bars indicate measurement error. (D,E) Interaction between H-ras and Gal-1 or the C-Raf-RBD (RBD) as indicated by the legend in (C) was determined by FLIM-FRET. Purified proteins as in (C) were incubated with fluorescent protein tagged proteins derived from BHK21 cell lysates (indicated with dotted cell outline). (D) Proteins from cell lysates were mRFP-tagged (red circle). (E) H-rasG12V labelled with mGFP (green circle) from lysates was used. Control is either mRFP-tagged C-Raf-RBD (upper column) or Gal-1 (lower column) incubated with 1 μM of CoA-488 in (D) and mGFP-H-rasG12V incubated with 1 μM of CoA-547 label in (E). (C–E) Binding of GTP-H-ras and the C-Raf-RBD served as a positive control. (D,E) Plotted values correspond to the mean ± SEM from three independent biological repeats. Numbers inside or above the bars indicate total number of imaged regions. The Methods section describes indicated statistical comparisons (ns, non-significant; ***p < 0.001).

Figure 2. H-rasG12V nanoclustering largely depends on effector interactions, while Gal-1 interacts with Raf-effectors.

(A) Electron microscopic nanoclustering analysis of mGFP-H-rasG12V and mGFP-H-rasG12V-D38A with or without antisense-mediated knockdown of Gal-1 in BHK21 cells. Normalized univariate K-functions, where maximal L(r)-r values above the 99% CI for complete spatial randomness indicate clustering at that value of r (number of membrane sheets analysed per condition, n ≥ 10). (B) Complexation between indicated mGFP-tagged H-ras mutants and mRFP-tagged C-Raf-RBD or Gal-1 was determined using FLIM-FRET in HEK293-EBNA cells transiently expressing above constructs (two independent biological repeats). (C) Complexation between indicated EGFP-tagged full-length Raf proteins and mRFP-tagged Gal-1 measured by FLIM-FRET in HEK293-EBNA cells (three independent biological repeats). Examples of FLIM-FRET images of cells, coexpressing indicated FRET-pairs or EGFP-tagged C-Raf-RBD as donor-only control. Image colour look-up table on the right shows fluorescence lifetimes. (B,C) Plotted values correspond to the mean ± SEM. Numbers inside and above the bars indicate total number of cells imaged. The Methods section describes the indicated statistical comparisons (***p < 0.001). Samples with coexpressed fluorescent proteins mGFP and mRFP (B), or EGFP and mRFP (C) served as FRET controls. Note that non-control sample FRET-values were all significantly different from the (FRET-)control sample. (D) Analysis of the interaction between endogenous Raf isoforms and Gal-1 in BHK21 cells using in situ proximity ligation assay (PLA). Representative confocal microscopy images of indicated proteins are shown. The sample with siRNA-mediated Gal-1 depletion served as a negative control. Cell nuclei were stained with DAPI. Red foci indicate positive signals for protein interactions and their quantification is shown in the graph. Scale bar is 21 μm.

Figure 3. Galectin-1 directly binds to the Ras binding domain of effectors.

(A) Interaction of Gal-1 with fragments of C-Raf (as can be derived from Supplementary Fig. 2A) or with the RBD of PI3Kα studied by FLIM-FRET in BHK21 cells, transiently expressing mCit-tagged Gal-1 and mRFP-tagged RBD-constructs (three independent biological repeats). Fluorescence lifetimes of FRET-samples were all significantly different from the donor-control. Plotted values correspond to the mean ± SEM. Numbers inside and above the bars indicate total number of cells imaged. The Methods section describes the indicated statistical comparisons (***p < 0.001). Samples with coexpressed fluorescent proteins mRFP and mCit served as FRET controls. Note that non-control sample FRET-values were all significantly different from the (FRET-)control sample. (B) GST pull-down experiments were performed by mixing bacterially purified Gal-1 with GST, GST-C-Raf-RBD or GST-PI3Kα-RBD immobilized on glutathione sepharose beads. GST was used as a negative control. Proteins retained on the beads were resolved by SDS-PAGE and Western blotted using a monoclonal antibody (M01) against Gal-1 for detection. (C) Corrected sensitized acceptor emission FRET data of 100 nM ATTO-488-labelled Gal-1 titrated with increasing concentrations of DY-547-labelled C-Raf-RBD (scheme on the left). Both proteins were purified from bacteria and labelled with the ACP-tag method. The dissociation constant (K_d) was determined from the shown curve fit on the dataset of Em_FRET as described in the Methods section. Plotted values correspond to the mean ± SEM.

Figure 4. Computational modelling and mutational validation of the Galectin-1/RBD complex.

(A) Computational representation of a monomer of Gal-1 (3W58) and C-Raf-RBD (1C1Y: herein RBD) complex from an optimized low energy molecular docking pose superimposed with dimeric Gal-1 from the same PDB deposition. Numbering of residues is according to sequences deposited in UniProt (P09382 – Gal-1_Homo sapiens, P04049 – C-Raf_Homo sapiens). The loop (loop 4) that undergoes major conformational and stereo-chemical changes between apo- and liganded Gal-1 is coloured orange (Supplementary Fig. 3A). Left: Note that the Gal-1 dimer interface, marked by the four mutated residues, is to the left near the N-terminus (N) of Gal-1. Residues forming the CBS are shown on the left monomer. The grey oval marks the region on the C-Raf-RBD that contacts Ras. Enlarged panel to the right shows a close-up view into the putative protein-protein interface and major interactions. Residues that were mutated in the RBD and showed an effect are marked with asterisks. The uncertainty regarding the interacting surface on Gal-1 is indicated by the translucent grey box. (B) Representative confocal images of HEK293-EBNA cells co-transfected with mGFP-Gal-1 and mRFP-C-Raf-RBD (RBD) mutated in the indicated residues. Columns represent imaged fluorescent channels, appropriate for the indicated construct. The nucleus is stained by DAPI. Overlay images show superposition of images to the left. Scale bar is 5 μm. (C) Interaction between mCit-Gal-1 (left) or mGFP-Gal-1 (right) and mRFP-tagged C-Raf-RBD and derived interfacial mutants studied using FLIM-FRET in HEK293-EBNA cells transiently expressing indicated constructs (three independent biological experiments). (D) Interaction between mGFP-H-rasG12V and mRFP-tagged C-Raf-RBD and derived interfacial mutants with or without coexpressed non-labelled Gal-1 (+Gal-1) studied using FLIM-FRET in HEK293-EBNA cells transiently expressing indicated constructs (three independent biological experiments). (C,D) Plotted values correspond to the mean ± SEM. Numbers inside the bars indicate total number of cells imaged. The Methods section describes indicated statistical comparisons (ns, non significant; *p < 0.05; **p < 0.01; ***p < 0.001). Samples with coexpressed fluorescent proteins mCit and mRFP (C) or mGFP and mRFP (C,D) served as a FRET control. Note that non-control sample FRET-values were all significantly different from the (FRET-)control sample.

Figure 5. A dimerization deficient Galectin-1 mutant loses its effect on Ras nanoclustering and signalling.

(A) Nanoclustering-FRET response of H-rasG12V in dependence of Gal-1 or its dimerization deficient mutant N-Gal-1 in HEK293-EBNA cells expressing mGFP-/mCherry-H-rasG12V. (B) RBD-recruitment FRET response of H-rasG12V in dependence of Gal-1 or its dimerization deficient mutant N-Gal-1 (mGFP-H-rasG12V and mRFP-C-Raf-RBD expressed in HEK293-EBNA) to assess effector translocation from the cytoplasm to active H-ras in plasma membrane nanoclusters. (C) Left, Western blot analysis of HEK293-EBNA lysates expressing mGFP-tagged H-ras and mRFP-tagged Gal-1 constructs as indicated. Serum-starved cells were stimulated with 100 ng/ml EGF for the indicated times. Total ERK and phospho-ERK (pERK) levels were then determined by immunoblotting. β-actin is the loading control. Right, Quantification of three independent repeats of Western blot data as shown on left. The pERK-signal was normalized to the total ERK-signal. (D) The nanoclustering-FRET response of H-rasG12V, N-rasG12V and K-rasG12V with increasing concentration of Gal-1. BHK21 cells were transiently co-transfected with mGFP-/and mCherry-tagged Ras constructs alone (1.0) or with antisense-Gal-1 (0.5) or non-labelled Gal-1 (3.4). The cellular total Gal-1 concentration relative to endogenous Gal-1 in control BHK21 cells ([Gal-1]rel.) is displayed to the left of the data. (E) The nanoclustering-FRET response of K-rasG12V. HEK293-EBNA cells transiently expressed mGFP-/mCherry-K-rasG12V and if indicated non-labelled Gal-1 or N-Gal-1. Note that in (D) the FRET-levels for K-rasG12V are lower than in (E), due to the higher Gal-1 level in BHK21 as compared to HEK293-EBNA cells (Supplementary Fig. 4E). (A,B,D,E) Plotted values correspond to the mean ± SEM of three independent biological experiments. Numbers inside the bars indicate total number of cells imaged. The Methods section describes the indicated statistical comparisons (ns, non significant; *p < 0.05; ***p < 0.001); comparisons in (D) were done against the 1.0 parent-control. Samples with coexpressed fluorescent proteins mGFP and mCherry (A,E) or mGFP and mRFP (B) served as a FRET control. Note that non-control sample FRET-values were all significantly different from the (FRET-)control sample.

Figure 6. Comparison of current and our new model for the mechanism of action of Gal-1 as a nanocluster scaffold.

(A) Current model. Direct interaction of active H-ras and Gal-1 stabilizes nanocluster. H-ras (depicted as yellow oval) activation supposedly makes the C-terminal farnesyl chain of H-ras more accessible for the prenyl-binding pocket of Gal-1 (depicted as a blue hexagon). This mechanistic step has not been described for other, similar trafficking chaperones, such as GDIs or PDEδ. Instead the spontaneous, activation state independent dissociation from the membrane is the basis for complexation by such chaperones in the cytoplasm. (B) In the new model proposed here, Raf effectors (depicted as violet rectangles) are recruited to active H-ras in nanoclusters on the plasma membrane. At higher concentrations Gal-1 can dimerize. Gal-1 binds directly to the Ras binding domain (RBD) of effectors, such as Raf. Thus, dimeric Gal-1 could stabilize effector (Raf)-dimers, which then act as the actual ‘scaffold’ for H-ras nanocluster. Note that effector and Gal-1 can form complexes already in the cytoplasm. Stacked dimers of H-ras + effector (Raf) + Gal-1 (box to right) would be nucleating the growth of H-ras nanocluster, a process that may be supported by the membrane environment.