Abstract
By developing a wide-field scheme for spectral measurement and implementing photoswitching, we synchronously obtained the fluorescence spectra and positions of ∼106 single molecules in labeled cells in minutes, which consequently enabled spectrally resolved, 'true-color' super-resolution microscopy. The method, called spectrally resolved stochastic optical reconstruction microscopy (SR-STORM), achieved cross-talk–free three-dimensional (3D) imaging for four dyes 10 nm apart in emission spectrum. Excellent resolution was obtained for every channel, and 3D localizations of all molecules were automatically aligned within one imaging path.
Similar content being viewed by others
References
Ha, T. Nat. Methods 11, 1015–1018 (2014).
Xie, X.S. & Trautman, J.K. Annu. Rev. Phys. Chem. 49, 441–480 (1998).
Moerner, W.E. & Orrit, M. Science 283, 1670–1676 (1999).
Rust, M.J., Bates, M. & Zhuang, X.W. Nat. Methods 3, 793–795 (2006).
Betzig, E. et al. Science 313, 1642–1645 (2006).
Hess, S.T., Girirajan, T.P.K. & Mason, M.D. Biophys. J. 91, 4258–4272 (2006).
Fölling, J. et al. Nat. Methods 5, 943–945 (2008).
Heilemann, M. et al. Angew. Chem. Int. Ed. Engl. 47, 6172–6176 (2008).
Trautman, J.K., Macklin, J.J., Brus, L.E. & Betzig, E. Nature 369, 40–42 (1994).
Macklin, J.J., Trautman, J.K., Harris, T.D. & Brus, L.E. Science 272, 255–258 (1996).
Lu, H.P. & Xie, X.S. Nature 385, 143–146 (1997).
Moerner, W.E. & Fromm, D.P. Rev. Sci. Instrum. 74, 3597–3619 (2003).
Howell, S.B. Handbook of CCD Astronomy 2nd edn. Ch. 6, 135–166 (Cambridge Univ. Press, 2006).
Xu, K., Babcock, H.P. & Zhuang, X. Nat. Methods 9, 185–188 (2012).
Bates, M., Huang, B., Dempsey, G.T. & Zhuang, X.W. Science 317, 1749–1753 (2007).
Shroff, H. et al. Proc. Natl. Acad. Sci. USA 104, 20308–20313 (2007).
Bossi, M. et al. Nano Lett. 8, 2463–2468 (2008).
Testa, I. et al. Biophys. J. 99, 2686–2694 (2010).
Huang, B., Wang, W.Q., Bates, M. & Zhuang, X.W. Science 319, 810–813 (2008).
Dempsey, G.T., Vaughan, J.C., Chen, K.H., Bates, M. & Zhuang, X. Nat. Methods 8, 1027–1036 (2011).
Sperlich, K. & Stolz, H. Meas. Sci. Technol. 25, 015502 (2014).
Acknowledgements
We thank S. Lee, A. Chiu and S. Moon for help in sample preparation and M. Wojcik for discussion. This work was partly supported by the College of Chemistry at UC Berkeley and the Lawrence Berkeley National Laboratory.
Author information
Authors and Affiliations
Contributions
Z.Z., S.J.K. and M.H. conducted the experiments. Z.Z., W.L. and K.X. analyzed the data. Z.Z., S.J.K. and K.X. prepared the manuscript with help from M.H. and W.L. K.X. conceived and supervised the project.
Corresponding author
Ethics declarations
Competing interests
The authors declare no competing financial interests.
Integrated supplementary information
Supplementary Figure 1 Spectral calibration of Path 2.
(a-d) Calibration using immobilized fluorescent beads and narrow bandpass filters. (a) Image of beads in the Image Mode. (b) Dispersed spectra in the Spectrum Mode. Magenta and green crosses mark the positions for 700 and 750 nm emission, respectively [from c,d]. (c,d) Spectra of the same beads with 700/10 nm and 750/10 nm band-pass filters. (e,f) Calibration using lasers: Images of a narrowed Slit 2 under illumination of 488, 560, and 647 nm lasers in the Image Mode and the Spectrum Mode, respectively. (g) Calibrated wavelength-pixel shift relationship. Results obtained from different calibration methods agreed well with each other. Black line is a third-order polynomial fit.
Supplementary Figure 2 Distribution of the spectral mean of single molecules across the camera.
(a) STORM image of Alexa Fluor 647-labeled microtubules (tubulin) in a fixed COS-7 cell. Magenta box corresponds to the same region as Fig. 1b,c. (b) Averaged single-molecule spectra for the 7 molecules in Fig. 1b,c (blue), 32 molecules across the same frame (green), 102,984 molecules for the same boxed region in 30 k frames (4.5 min; magenta), and 573,527 molecules across the camera in 30 k frames (black). (c) Statistics of the measured single-molecule spectral means for all molecules across the camera (red) and molecules in different divided areas (as labeled in a) of the camera (black). Error bars represent standard deviation between single molecules. Grey band corresponds to the distribution of all molecules.
Supplementary Figure 3 Single-molecule spectra for Alexa Fluor 647 labeled to different cellular targets.
(a) STORM image of Alexa Fluor 647-labeled microtubules (tubulin) and outer membrane of mitochondria (TOM20) in a fixed COS-7 cell. (b–d) Representative single-molecule spectra obtained on different targets: microtubules (b), outer membrane of mitochondria (c), and nonspecific labeling (d), from regions marked by the orange, magenta, and cyan boxes in a, respectively. (e) Averaged single-molecule spectra over the three selected regions. (f–h), Distribution of spectral mean for single molecules detected in the three selected regions.
Supplementary Figure 4 Averaged single-molecule spectra of different dyes.
Minor irregular features are attributed to inadequate correction for the wavelength-dependent variation in detector responsivity (Online Methods).
Supplementary Figure 5 The separated four dye channels for the full-frame image of Figure 2.
(a) Dyomics 634. (b) DyLight 650. (c) CF660C. (d) CF680. Scale bars: 2 µm.
Supplementary Figure 6 Identification/categorization of single molecules based on the measured single-molecule spectra for Dyomics 634, DyLight 650, CF660, and CF680.
Four cell samples were each immunofluorescence-labeled with one of the four dyes, and ~106 single-molecule spectra were measured in each sample over the course of a few minutes. Each detected single molecule was categorized as one of the four dyes based on a comparison of its measured spectrum and the predetermined, averaged single-molecule spectra of the four dyes (Supplementary Fig. 4). (a) Fraction of single molecules that were identified as each dye for the four dye-labeled samples. (b) A log scale presentation of a. Asterisks mark channels with fractions lower than 10−3. Crosstalk into incorrect color channels is <2% between all channels (highest being 1.9% for labeled CF680 single molecules being misidentified as CF660C).
Supplementary Figure 7 The separated four dye channels of the full-frame image for the data shown in Figure 3a–e.
The z positions are color-coded on the same scale for all channels (violet: closest to substrate; red: farthest away). Scale bars: 2 µm.
Supplementary Figure 8 3D localization distribution for single molecules of different dyes.
Clusters of localizations, due to either multiple dye molecules tagged to the same antibody or the repetitive activation of the same dye molecule, were aligned by their center of mass to generate the 3D presentation of the localization distribution18,22. Localizations from >100 clusters (each containing >9 localizations) are summarized for each dye channel from the dataset shown in Supplementary Fig. 7. Histograms of distribution in x, y, z are fitted to Gaussian functions (red curves), and the resultant standard deviations (σX, σY, σZ) are shown in each plot. (a) Dyomics 634. (b) DyLight 650. (c) CF660C. (d) CF680.
Supplementary Figure 9 Consecutive virtual vertical yz sections (each 150 nm in x) for an area in Figure 3a.
Three of the virtual vertical yz-sections are presented in Fig. 3e.
Supplementary Figure 10 Correction for the wavelength-dependent variation in the responsivity of the detection system.
(a) Measured spectrum of a broadband quartz tungsten-halogen lamp (black solid line), and the obtained wavelength-dependent correction factor (red dotted line) after subtraction of a smoothed curve. (b-g) Measured raw spectral data without application of the wavelength-dependent correction factor. Similar irregular features are observed for different dyes in the uncorrected raw spectra, e.g., dips at ~690 nm and ~720 nm, in agreement with the drop in responsivity at corresponding wavelengths.
Supplementary information
Supplementary Text and Figures
Supplementary Figures 1–10 (PDF 12967 kb)
The simultaneously recorded single-molecule images (via Path 1) and single-molecule spectra (via Path 2) obtained within 1 s for the full CCD frame of the sample shown in Fig. 1b,c and Supplementary Fig. 2
The EM-CCD camera recorded at 110 frames per second (FPS), and so the playback is slowed down by 4.4-fold when played at the standard video frame rate of 25 FPS. Yellow, magenta, and green crosses respectively mark the mapped spectral positions of 647, 700 and 750 nm for each molecule. White boxes correspond to the region presented in Fig. 1b,c (first frame). Scale bar: 2 µm. (AVI 19090 kb)
Rights and permissions
About this article
Cite this article
Zhang, Z., Kenny, S., Hauser, M. et al. Ultrahigh-throughput single-molecule spectroscopy and spectrally resolved super-resolution microscopy. Nat Methods 12, 935–938 (2015). https://doi.org/10.1038/nmeth.3528
Received:
Accepted:
Published:
Issue Date:
DOI: https://doi.org/10.1038/nmeth.3528
- Springer Nature America, Inc.
This article is cited by
-
Spectroscopic single-molecule localization microscopy: applications and prospective
Nano Convergence (2023)
-
Event-based vision sensor for fast and dense single-molecule localization microscopy
Nature Photonics (2023)
-
Combining pMINFLUX, graphene energy transfer and DNA-PAINT for nanometer precise 3D super-resolution microscopy
Light: Science & Applications (2023)
-
Tetra-color superresolution microscopy based on excitation spectral demixing
Light: Science & Applications (2023)
-
Global fitting for high-accuracy multi-channel single-molecule localization
Nature Communications (2022)