Fast and high resolution single-cell BRET imaging

Abstract

Resonance Energy Transfer (RET)-based technologies are used to report protein-protein interactions in living cells. Among them, Bioluminescence-initiated RET (BRET) provides excellent sensitivity but the low light intensity intrinsic to the bioluminescent process hampers its use for the localization of protein complexes at the sub-cellular level. Herein we have characterized the methodological conditions required to reliably perform single-cell BRET imaging using an extremely bright luciferase, Nanoluciferase (Nluc). With this, we achieved an unprecedented performance in the field of protein-protein interaction imaging in terms of temporal and spatial resolution, duration of signal stability, signal sensitivity and dynamic range. As proof-of-principle, an Nluc-containing BRET-based sensor of ERK activity enabled the detection of subtle, transient and localized variations in ERK activity in neuronal dendritic spines, induced by the activation of endogenous synaptic NMDA receptors. This development will improve our comprehension of both the spatio-temporal dynamics of protein-protein interactions and the activation patterns of specific signaling pathways.

Figure 1. Experimental protocol scheme.

(a) Cells were transfected with the donor entity and a red-fluorescent transfection reporter, DsRed (left) or the BRET-positive fusion (right). These two populations of transfected cells were subsequently mixed (24 hours after transfection). The following illustrations (b–e) were obtained with Nluc as BRET donor. (b) Cells expressing the donor, Nluc (+DsRed), or the donor fused to the acceptor, Nluc-Venus, could be identified by direct fluorescence excitation (GFP and DsRed merged image) when studied in the same microscopic field, allowing a direct comparison of cells that did or did not display BRET. In the absence of light excitation, Em480 (c) and Em535 (d) images were acquired in the presence of the relevant luciferase substrate. (e) The pixel-by-pixel ratio of images obtained at 535 nm, compared to those obtained at 480 nm, demonstrates a strong signal in cells expressing the BRET-positive fusion (cells 1, 2 and 3), but only a marginal signal in cells expressing the donor alone (cells 4 and 5). As a control it must be noted that some cells from the DsRed + luciferase-transfected population which expressed only luciferase displayed the same 535 nm/480 nm ratio as cells co-expressing DsRed with luciferase.

Figure 2. The time resolution of BRET imaging is 10-times greater with Nluc compared to Rluc8.

(a) HEK cells were co-transfected with either the BRET fusions Nluc-Venus (top) or Rluc8-Venus (bottom), or the donor entities Nluc (top) or Rluc8 (bottom), together with the DsRed transfection reporter. On the left, GFP and DsRed fluorescence (merged images) was used to discriminate between cells expressing the BRET fusion with those expressing the donor entity with DsRed. On the right, the 535 nm/480 nm ratio image was obtained from either a 1 s or 5 s sequential acquisition per channel, 5 min after the addition of the relevant substrate, Furimazine (top) or Coelenterazine H (bottom). Square areas are shown at a higher magnification in the right panels. Scale bars: white = 20 μm; black = 5 μm. (b) Em480 and (c) Em535 images obtained for acquisition times ranging from 50 ms to 10 s were recorded 5 min after the addition of Coelenterazine H or Furimazine. (d) Em480 and (e) Em535 intensities recorded for the indicated acquisition times along a cross-section of cells 1 (left) and 2 (right) in A. (f) Average Em480, Em535, and 535 nm/480 nm signal intensities, expressed as a function of acquisition time per channel for Nluc-Venus or Rluc8-Venus. Acquisition times over 1 s per channel were required to detect reliable BRET signals with Rluc8 while an acquisition time of 100 ms was sufficient with Nluc. (g) Em480 signal/background ratio absolute values (left), or normalized to their maximal value (right). Nluc produced a higher signal/background ratio than Rluc8. A 1.7 s acquisition time was required to reach 50% of the signal/background ratio with Nluc-Venus versus 3.2 s for Rluc8-Venus. (f, g) Each point of the graph represents the mean ± SEM obtained from 5 to 7 cells of similar fusion expression levels (YFP intensities: Nluc-Venus = 745 ± 28 RLUs; Rluc8-Venus = 771 ± 33 RLUs), from 3 square areas per cell.

Figure 3. Nluc generates a stable BRET signal over time in single-cell imaging.

(a) Merged images of YFP and DsRed fluorescence were used to discriminate between cells expressing the BRET fusions Nluc-Venus (top) or Rluc8-Venus (bottom) (cells 1, 2 and 3), and cells expressing the donor + DsRed (cells 4 and 5) or the donor alone (cell 6). Em480, Em535 and the 535 nm/480 nm pseudo-colored ratio images were recorded with 1 s (top) or 5 s sequential acquisitions per channel (bottom) after 5 min incubation with the relevant substrate. Square areas are shown at a higher magnification in the right panels. (b,c) Em480, Em535 and the 535 nm/480 nm ratio images were acquired every 5 min from 0 to 60 min after substrate incubation with cells expressing Nluc and Nluc-Venus (b) or Rluc8 and Rluc8-Venus (c). (d,e) The intensity of Em480, Em535, and 535 nm/480 nm ratio signals were measured every minute for 60 min with Nluc (d) and Rluc8 (e) as the energy donors within a square area of cells numbered in (a). White scale bar = 20 μm; Black scale bar = 5 μm.

Figure 4. Nluc improves the sensitivity and dynamic range of BRET measurements in single-cell imaging.

(a) Bar graph of the average 535 nm/480 nm ratio intensities of Rluc8, Rluc8-Venus, Nluc and Nluc-Venus obtained from 5 s (Rluc8) or 1 s (Nluc) sequential acquisitions per channel between 5 and 20 min after substrate addition. Nluc decreased the basal 535 nm/480 nm ratio. Nluc triggered a more efficient energy transfer with Venus compared to Rluc8. (b) Average of the net 535 nm/480 nm ratio intensities for Rluc8-Venus and Nluc-Venus calculated by subtracting the basal 535 nm/480 nm ratio obtained with Rluc8 and Nluc from the 535 nm/480 nm ratio of Rluc8-Venus and Nluc-Venus respectively. The dynamic window for 535 nm/480 nm ratio measurements was significantly better using Nluc compared to Rluc8. (c) Average 535 nm/480 nm ratio intensity measured every minute for 60 min after substrate addition in the square area shown in cells expressing Rluc8-Venus (cell 1) and Nluc-Venus (cell 3) in Fig. 3a. Close-up of the 5 to 15 min time period after substrate addition when the 535 nm/480 nm ratio was stable. (d) Mean standard deviation of the 535 nm/480 nm ratios between 5 and 20 min after substrate incubation, expressed as a percentage of the mean ratio signal. The temporal noise was significantly lower with Nluc than Rluc8. Bars represent mean ± SEM obtained from 8 or 9 cells. Mann & Whitney statistical analysis: **p < 0.01, ****p < 0.0001.

Figure 5. The Nluc-containing ERK-activity reporter improves the sensitivity and spatio-temporal resolution of BRET signals.

(a,b) Schematic representations of the conformational changes of the REV (a) and YEN (b) sensors induced upon ERK activation. Adapted from Xu et al.. (c,d) Average 535 nm/480 nm ratio intensity measured every minute for 30 min after PMA (PMA) or buffer (control) addition to REV- (c) and YEN- (d) transfected cells. 535 nm/480 nm ratio images obtained before (CT) and after 10 min PMA application (PMA (600 sec)). (e) Mean standard deviation of the 535 nm/480 nm ratios in time (left) and space (right), expressed as a percentage of the mean ratio signal. Left- The temporal noise was measured from 1200 to 1620 s after PMA incubation. Right- The spatial noise was measured 600 s after PMA application within a square area of 21 × 21 pixels. Bars are mean ± SEM obtained from 6 cells. Mann & Whitney statistical analysis: *p < 0.05.

Figure 6. The Nluc-containing reporter is able to detect subtle ERK-activity modulations in the dendritic spines of neurons.

(a,b) The intensities of the 535 nm/480 nm ratio over time, recorded in the soma or dendrites of REV- (a) or YEN- (b) transfected neurons, stimulated or not (control) with Glycine. One representative image illustrates the 535 nm/480 nm ratio at the indicated times. Each point of the graphs represents the mean ± SEM for 3 to 6 neurons and 7 regions per neuron, for each time point. (c) The 535 nm/480 nm ratio of individual spine (arrowhead) and shaft (scare) areas were measured over time. Bold traces represent lines of best fit from 12 traces. One representative image illustrates the 535 nm/480 nm ratio at the indicated times. (d) The intensity (Y axis) and time position (X axis) of the maximal 535 nm/480 nm ratios measured in 12 shaft areas (upper panel) and 12 spines (bottom panel) in the absence and presence of glycine. Each plot is the mean ± SEM of the time positions and mean ± SEM of the peak intensities. Mann & Whitney statistical analysis: *p < 0.05. Glycine stimulation induced a significant increase of the maximal BRET intensity in spines, compared to non-stimulated neurons.