Super-resolution mapping of glutamate receptors in C. elegans by confocal correlated PALM

Abstract

Photoactivated localization microscopy (PALM) is a super-resolution imaging technique based on the detection and subsequent localization of single fluorescent molecules. PALM is therefore a powerful tool in resolving structures and putative interactions of biomolecules at the ultimate analytical detection limit. However, its limited imaging depth restricts PALM mostly to in vitro applications. Considering the additional need for anatomical context when imaging a multicellular organism, these limitations render the use of PALM in whole animals difficult. Here we integrated PALM with confocal microscopy for correlated imaging of the C. elegans nervous system, a technique we termed confocal correlated PALM (ccPALM). The neurons, lying below several tissue layers, could be visualized up to 10 μm deep inside the animal. By ccPALM, we visualized ionotropic glutamate receptor distributions in C. elegans with an accuracy of 20 nm, revealing super-resolution structure of receptor clusters that we mapped onto annotated neurons in the animal. Pivotal to our results was the TIRF-independent detection of single molecules, achieved by genetic regulation of labeled receptor expression and localization to effectively reduce the background fluorescence. By correlating PALM with confocal microscopy, this platform enables dissecting biological structures with single molecule resolution in the physiologically relevant context of whole animals.

Figure 1. Labeling strategy for confocal correlated PALM.

(a) Schematic overview of the C. elegans nervous system, showing the major neuronal ganglia and processes. Ganglia containing GLR-1 expressing neurons are marked in magenta. L1 larvae are approximately 250 μm long and 25 μm in diameter, but grow to 1.3 mm long and 80 μm diameter when reaching the adult stage. (b) Genomic position of the glr-1-gene on chromosome III. Boxes represent the exons, while the connecting bars represent the introns. Two amplicons were obtained by genomic PCR: the first containing the glr-1 gene with approximately 4 kb of putative promoter sequence upstream from the start codon, and the second containing only the putative promoter region. The former was fused to the sequence encoding mEOS2 while the latter was placed in front of the eGFP-encoding sequence. (c) Confocal image showing a partial Z-projection of the right lateral side of transgenic C. elegans expressing both the Pglr-1::glr-1::mEOS2 and the Pglr-1::eGFP constructs. The GLR-1 expressing neurons are represented in cyan, while the localization of the GLR-1-mEOS2 fusion protein is coded in magenta. Annotated neurons indicated by white arrows. Scale bar measures 2 μm.

Figure 2. Single molecule detection in C. elegans.

(a,b) Raw images (10 by 10 pixels) data showing 2 randomly chosen single molecules inside C. elegans. (c,d) 3D representation of the single molecule signal (top), 2D Gaussian fit (middle) and residuals after fitting (bottom) shown in (a) and (b) respectively. The y-axes on the 3D-plots denote the intensity in arbitrary units (e,f) Intensity time trace of the molecules shown in (a,b) indicating that mEOS2 blinks on and off multiple times, leading to possible multiple detections. (g) Graph showing the amount of detected unique molecules in function of the allowed dark time (t_d) between detections (red circles). Exponential fit (blue line) shows the onset of a plateau at around 2 seconds.

Figure 3. PALM of C. elegans VNC.

(a,b) Head region of C. elegans with GLR-1 expressing neurons labeled with eGFP (blue) and GLR-1 molecules labeled with mEOS2 (red). eGFP fluorescence was recorded in standard epi-fluorescence mode and only clearly outlines the ventral nerve cord (VNC). Single GLR-1 molecules detected by PALM imaging were plotted as spots with a width of 214 nm to simulate diffraction-limited microscopy. When zooming in (b), GLR-1 puncta are visible in the VNC. (c) PALM image plotted at 25 ± 0.2 nm average resolution ± standard deviation, similar to (b), showing a more detailed structured of the GLR-1 puncta in the VNC. Scale bars measure 1 μm (a) or 250 nm (b,c). (d) Cluster thickness plotted versus cluster length (red dots). The function y = x is shown in blue, the function y = 3x is shown in green. Most clusters are situated between these lines. (e) Cluster thickness plotted as a function of number of molecules per cluster and (f) cluster length plotted as a function of number of molecules per cluster. Distributions shown (e,f) are fitted with a logarithmic function (blue curve). Function and fit coefficients ± s.d. are shown in box.

Figure 4. Confocal correlated PALM in C. elegans.

(a) Partial confocal Z-projection (15 optical slices, total thickness of 5.30 μm) of eGFP fluorescence marking GLR-1 positive neurons and their processes (blue) overlaid with transmission image of the C. elegans head region (grey). White arrowheads indicate the nerve ring and the VNC. (b) Enlargement of GLR-1 expressing head neurons from Z-projection in (a). (c,d) Close up of the AVEL neuron (c) and the nerve ring (d) indicated in (b) with the distribution of GLR-1 mapped onto the neuron by ccPALM. (e,f) Close up of box in panels (c) and (d), respectively. Scale bars indicate 2 μm (a,b) 250 nm (c,d), or 100 nm (e,f). Cut-off resolution of PALM images (c–f) is 20 nm.