Super-resolution imaging of ciliary microdomains in isolated olfactory sensory neurons using a custom two-color stimulated emission depletion microscope

Abstract

We performed stimulated emission depletion (STED) imaging of isolated olfactory sensory neurons (OSNs) using a custom-built microscope. The STED microscope uses a single pulsed laser to excite two separate fluorophores, Atto 590 and Atto 647N. A gated timing circuit combined with temporal interleaving of the different color excitation/STED laser pulses filters the two channel detection and greatly minimizes crosstalk. We quantified the instrument resolution to be ∼81 and ∼44 nm, for the Atto 590 and Atto 647N channels. The spatial separation between the two channels was measured to be under 10 nm, well below the resolution limit. The custom-STED microscope is incorporated onto a commercial research microscope allowing brightfield, differential interference contrast, and epifluorescence imaging on the same field of view. We performed immunolabeling of OSNs in mice to image localization of ciliary membrane proteins involved in olfactory transduction. We imaged Ca2+-permeable cyclic nucleotide gated (CNG) channel (Atto 594) and adenylyl cyclase type III (ACIII) (Atto 647N) in distinct cilia. STED imaging resolved well-separated subdiffraction limited clusters for each protein. We quantified the size of each cluster to have a mean value of 88±48 nm and 124±43 nm, for CNG and ACIII, respectively. STED imaging showed separated clusters that were not resolvable in confocal images.

Fig. 1

(a) Absorption and emission spectra of Atto 590 and Atto 647N, the two colors used in the STED, are shown, along with the excitation and STED laser wavelengths used. Emission detection bands shown are 600 to 640 nm for Atto 590 and 660 to 700 nm for Atto 647N. (b) Interleaved excitation/STED pulses for each color in the time domain and an illustration of the temporal gated collection are shown. Green boxes indicate the gates for the Atto 590 signal and the red boxes indicate the gates for the Atto 647N signal. The output signal from each gating circuit is sent to the computer.

Fig. 2

Optical schematic for the two-color STED microscope. Each beam is bandpass filtered and then sent through a half-wave plate and polarizer in succession, allowing clean polarization and individual power control. Excitation beams are combined into one polarization maintaining (PM) fiber and the STED beams are combined into another PM fiber using dichroic mirrors. STED beams pass through a vortex phase plate to generate a donut shape at the focus of the objective. A polarizing beamsplitter combines the beams and a quarter-wave plate makes the polarization circular. The beams are sent through a custom dichroic, which passes the four STED/excitation laser beams and reflects the fluorescence emission. The two emission bands are then separated using another dichroic beamsplitter. Fluorescence is focused into a multimode fiber that is coupled to an APD. The signal from the APDs goes through the custom gating circuit to counter inputs on the data acquisition (DAQ) board in the computer. The same DAQ board also controls the scanning of the piezo stage to acquire an image.

Fig. 3

Confocal and STED images of $\sim 45\mathrm{nm}$ fluorescent beads to determine resolution and spatial overlap of the two channels. (a) Red beads imaged in the Atto 590 channel, (b) dark red beads imaged in the Atto 647N channel, (frame size: $1\mu \mathrm{m}\times 1\mu \mathrm{m}$). (c) Confocal and (d) STED images of custom double-labeled beads taken simultaneously in the Atto 590 channel (green) and Atto 647N channel (red), ($\text{scale bar}=1\mu \mathrm{m}$). Right: Boxcar plots for the fitted FWHM of the beads in the (e) Atto 647N and (f) Atto590 channels indicate increased resolution obtained with STED over confocal imaging. A least squares fit to a 2-D-Gaussian function was performed to calculate the FWHM for individual beads shows: $218\pm 26\mathrm{nm}$ (confocal) and $93\pm 14\mathrm{nm}$ (STED) for Atto 590 and $347\pm 36\mathrm{nm}$ (confocal) and $63\pm 5.6\mathrm{nm}$ (STED) for the Atto 647N channel, $N=87$ beads (STED), $N=36$ beads (confocal). The average separation between the fitted centers of the Gaussian in the two channels was $\mathrm{\Delta }x=3.4\pm 7.4\mathrm{nm}$ and $\mathrm{\Delta }y=5.4\pm 6.5\mathrm{nm}$. Two-color image pixel size: 19.5 nm, pixel dwell time: 50 microseconds and total imaging time: 0:38.

Fig. 4

Effects of cross-talk rejection using gating circuit for two-color STED. Samples of either red or dark red 40 nm fluorescent beads were imaged in both color channels with or without gating. Image histograms are reported by summing the intensity over ROI drawn around each individual bead. Green (dashed): Atto 590 channel. Red: Atto 647N channel. The gating circuit greatly reduces the background signal without any reduction in counts in the correct channel.

Fig. 5

Canonical transduction pathway in the cilia of OSNs. An odor ligand binding to a specific receptor activates the G-protein ${\mathrm{G}}_{\mathrm{olf}}$, which then activates ACIII to convert intracellular ATP to cAMP. The CNG channel is opened by binding nearby cAMP and allows depolarizing cations, including ${\mathrm{Ca}}^{2+}$, to enter the cell. High ${\mathrm{Ca}}^{2+}$ concentration may also activate downstream ${\mathrm{Cl}}_{\mathrm{Ca}}$ channels, which further depolarizes the cell. Also illustrated are representations of the antibodies conjugated with Atto 594 and Atto 647N that label the CNG and ACIII proteins.

Fig. 6

Two-color STED images of OSNs labeled for ACIII, CNGA2, and acetylated tubulin. Panel (a) and (d) show the widefield epifluorescence images of the OSN labeled with Alexa Fluor 488 against acetylated-tubulin, while panel (b),(e) and (c),(f) show the two-color images in confocal and STED, respectively, with green indicating Atto 594-labeled CNGA2 and red indicating Atto 647N-labeled ACIII. (g) Boxcar plots for clusters on OSN cilia, combining clusters on both OSNs. Top: Atto 594 channel STED enhancement, showing the decrease in imaged cluster FWHM in going from confocal ($413\pm 150\mathrm{nm}$, $N=49$) to STED ($88\pm 48\mathrm{nm}$, $N=86$). Bottom: Atto 647N channel STED enhancement, confocal ($591\pm 160\mathrm{nm}$, $N=37$) and STED ($124\pm 43\mathrm{nm}$, $N=118$). The FWHM was calculated from a rotated 2-D Gaussian fit to $21\times 21\text{\hspace{0.17em}\hspace{0.17em}pixels}$ ($9.8\mathrm{nm}/\text{pixel}$) ROIs that were centered on the peaks of clusters. All images taken with pixel size: 9.8 nm, pixel dwell time: 30 microseconds and total imaging time: 3:56.