Mapping of DDR1 distribution and oligomerization on the cell surface by FRET microscopy

Abstract

Activation of discoidin domain receptor (DDR) 1 by collagen is reported to regulate cell migration and survival processes. While the oligomeric state of DDR1 is reported to play a significant role in collagen binding, not much is known about the effect of collagen binding on DDR1 oligomerization and cellular distribution. Using fluorescence resonance energy transfer (FRET) microscopy, we monitored the interaction between DDR1 tagged with cyan fluorescent protein and DDR1 tagged with yellow fluorescent protein in live cells. Significant FRET signal indicative of receptor dimerization was found even in the absence of collagen stimulation. Collagen stimulation induced aggregation of DDR1, followed by a sharp increase in FRET signal, localized in the regions of aggregated receptor. Further analysis of DDR1 aggregation revealed that DDR1 undergoes cytoplasmic internalization and incorporation into the early endosome. We found the kinetics of DDR1 internalization to be fast, with a significant percentage of the receptor population being internalized in the first few minutes of collagen stimulation. Our results indicate that collagen stimulation induces the aggregation and internalization of DDR1 dimers at timescales much before receptor activation. These findings provide new insights into the cellular redistribution of DDR1 following its interaction with collagen type I.

Fig. 1

The fluorescent DDR1 constructs are functional. (a) Schematic representation of the DDR1–YFP fusion proteins. (b) HEK293 cells, native or transiently transfected with DDR1–YFP, were stimulated with 10 μg/ml collagen type I for 90 min (as indicated), and cell lysates were imunoprecipitated with anti-DDR1 antibodies. Following SDS-PAGE and Western blot analysis, the membranes were probed with anti-phosphotyrosine antibodies (upper blot) or anti-DDR1 antibodies (lower blot). Collagen stimulation increases the phosphorylation level of fluorescently labeled DDRs.

Fig. 2

FRET analysis of DDR1 dimerization. 3T3 cells transiently transfected with both DDR1–YFP and DDR1–CFP were imaged live. For the same region, images were acquired sequentially in the I_AA, I_DD, and I_DA channels (a–c); the corrected and normalized FRET signal F_C/A is displayed in pseudocolor (d). Positive FRET signal indicative of dimmer formation is recorded over the entire cell surface. Regions that display higher signal intensities were found to be associated with variations in the acceptor–donor ratio (r). (Inset, d) For such a selected ROI, the FRET index, receptor levels, and 1/(1+r) values were plotted alongside. All scale bars represent 20 μm.

Fig. 3

The FRET index was plotted against the acceptor levels for nonstimulated samples. All data points (ROIs) included have similar A/D ratios, and the FRET index shows little dependence on the receptor levels.

Fig. 4

Collagen induces DDR1 aggregation and increase in FRET signal. 3T3 cells cotransfected with DDR1–YFP and DDR1–CFP were imaged before and after stimulation with collagen type I (as indicated); upper row shows images in the YFP channel (I_AA) demonstrating uniform receptor distribution in the nonstimulated sample (a); collagen stimulation induces aggregation of DDR1 (c); lower row (b, d) shows FRET analysis of the corresponding images in the upper row. The stimulated sample (d) exhibits higher FRET values in the regions of aggregated DDR1—as compared with similar regions in the nonstimulated sample (b). The ROI indicated by white squares in the upper row images were magnified in both sets of images for clear demonstration of receptor aggregation and increase in FRET. All scale bars represent 20 μm.

Fig. 5

The FRET index was plotted against the acceptor levels for collagen-stimulated samples (5-min time point). All data points (ROIs) included have similar A/D ratios, and the FRET index shows little dependence on the receptor levels.

Fig. 6

Dynamics of DDR1 aggregation. 3T3 cells transiently transfected with DDR1–YFP were imaged live before and after stimulation with collagen type I. (a) Nonstimulated cells show uniform distribution of the receptor. (b) After 5 min of collagen stimulation, DDR1 aggregates in clusters. (c) Increased stimulation time results in redistribution of a majority of DDR1 aggregates. Intensity levels in the nonaggregated regions revert to values similar to those observed before collagen stimulation.

Fig. 7

Quantitative analysis of the aggregation process depicted in Fig. 6. Aggregation percentage is depicted in green (left axis), together with the increase in FRET (blue; right axis). The two processes are shown to have similar dynamics in the initial phase, with a sharp increase immediately following collagen stimulation and little variation thereafter. Later time points (60 min) show redistribution of the DDR1 aggregates, while the FRET index in the remaining aggregates is similar to that recorded at earlier stimulation points.

Fig. 8

DDR1 aggregation takes place through receptor internalization. HEK293 cells were cultured on glass-bottom culture dishes, transiently transfected with DDR1–YFP, and stimulated with collagen type I for 10 min (as indicated). Following stimulation, cells were imaged live using wide-field fluorescence microscopy (a and b). Nonstimulated sample (a) shows localization of the receptor on the cell membrane. (b) Stimulated sample shows loss of fluorescence from the membrane and aggregation of the receptors. Confocal imaging of fixed cells confirms localization of the receptor on the cell membrane in the nonstimulated sample (c), while stimulated cells (d) show a loss of receptor signal from the membrane accompanied by aggregation inside the cytoplasm. Nuclear staining is shown in blue. All scale bars represent 10 μm.

Fig. 9

Collagen stimulation decreases the amount of soluble DDR1 in the cell lysate. (a) HEK293 cells, transiently transfected with DDR1–YFP, were stimulated with 10 μg/ml collagen type I for 0-min, 10-min, 30-min, and 60-min intervals (as indicated). Cell lysates were run on SDS-PAGE and, following Western blot analysis, the membranes were probed with anti-DDR1 antibodies (upper blot); DDR1–YFP is indicated by the horizontal arrow. A control untransfected sample (rightmost lane) does not show the YFP-tagged DDR1. The membranes were reprobed with anti-GAPDH antibodies to verify protein expression and loading (lower blot). (b) The signal in the DDR1 bands was quantified and normalized to the signal in the corresponding GAPDH bands. The signal in the collagen-stimulated bands is given as the percentage of the signal in the nonstimulated (0 min) band.

Fig. 10

DDR1 aggregates colocalize with Rab5a. HEK293 cells were transiently transfected with both DDR1–CFP and Rab5a-GFP and stimulated with collagen type I for 15 min (as indicated). Confocal images of nonstimulated cells are shown on the upper row (a–c), and stimulated cells are shown on the lower row (d–f). CFP, red; GFP, green; overlap of the two, yellow. (f) The aggregated DDR1 (indicated by white arrows) in the stimulated cells overlaps with Rab5a. All scale bars represent 10 μm.

Fig. 11

Object-based colocalization analysis. (a) Longitudinal (L) and transversal (T) linear ROIs along a DDR1–CFP aggregate. Intensity profiles along the longitudinal (b) and transversal (c) ROIs in the GFP (green) and CFP (red) channels overlap, demonstrating colocalization of DDR1–CFP and Rab5a-GFP.

Scheme 1

Proposed model of DDR1 activation mechanism. DDR1 exists on the cell membrane as a mixture of monomers and dimers. High-affinity interaction between DDR1 and collagen monomers induces internalization of DDR1 dimers and activation of the endosomal signaling pathway. Endosomal signaling induces DDR1 activation, followed by recycling of DDR1 dimers to the cell surface.