Multiplexed Super-Resolution Imaging with Programmable DNA Probes

  • Fig. 1: Stochastic reconstruction microscopy and DNA-PAINT. (a) Stochastic super-resolution microscopy. Top-to-bottom: Schematic of dye positions alongside with diffraction-limited representation. Stochastic activation of a small subset of dyes allows precise nanometer-scale localization in each acquisition frame. The final reconstructed image has sub-diffraction resolution. (b) DNA-PAINT concept. Dye-labeled imager strands transiently bind to docking sites thus producing an apparent blinking under TIR illumination [6]. (c) Nanoscale DNA origami structures are “labeled” with docking strands resembling digit “0”. The comparison between the diffraction-limited and the DNA-PAINT image highlights the increase in obtained imaging resolution [8].
  • Fig. 2: Exchange-PAINT concept. (a) Multiple targets are labeled with orthogonal docking sequences (S1* to S4*). Imaging is performed sequentially using one imager strand species (S1 to S4) at a time. Each imaging round is followed by a brief washing step (~1–2 min). After image acquisition for all rounds is completed, pseudocolors for each round are assigned and the final overlay image is constructed. (b) Four rounds of Exchange-PAINT imaging reveals docking strand patterns representing digits 0–3 on a single DNA origami structure.

Super-resolution microscopy has changed the way we look at biology by allowing imaging below the diffraction-limit of light. However, most implementations require specialized equipment or buffer conditions and multiplexed imaging remains challenging. We use the transient binding of dye-labeled oligonucleotides (DNA-PAINT) for straightforward super-resolution imaging. Additionally, the programmability of DNA-based probes enables a powerful and easy-to-implement multiplexing approach (Exchange-PAINT).

Fluorescence microscopy is an invaluable characterization tool in modern biological research; It enables the observation of multiple molecular species (e.g. different proteins) with high specificity by using spectrally distinct dyes as labeling probes.

Although single dye molecules are only ~1 nm in size, they appear as "broadened" ~200 nm spots when imaged using diffraction-limited optics in a standard fluorescence microscope. Therefore, two point-like objects, such as two dyes, cannot be distinguished anymore if they are spaced closer than 200 nm. However, cellular biological processes ultimately happen on the level of single biomolecules on the nanometer scale. Thus the goal of optical microscopy is to be able to visualize these processes with sub-diffraction spatial resolution.

Pursuing the quest for higher spatial resolution, far-field fluorescence microscopy underwent a true renaissance in recent years by the invention of methods circumventing the classical diffraction limit, i.e. super-resolution microscopy [1].

In most implementations dyes are "switched" between fluorescence bright and dark states, so that individual molecules can be localized consecutively. This can be obtained with two strategies: "Targeted" switching limits the effective fluorescence excitation area to below the diffraction limit (e.g. STED [2]); "Stochastic" switching temporally separates fluorescence emission from single molecules by switching them from dark to bright states and back. Here, photoswitchable proteins (PALM [3]) or organic dyes (STORM [4]) are used as probes. During the acquisition of each frame in a stochastic reconstruction method, most molecules are switched to a dark state; very few molecules are switched to a bright state and are localized with nanometer precision by fitting their emission with a Gaussian function.

This procedure is illustrated in Figure 1a.

Although current super-resolution techniques are transforming the way we look at biology today, they are rather complex to implement and multiplexed visualization of different molecules is challenging. To enable easy-to-implement super-resolution imaging, DNA-PAINT (a variation of point accumulation for imaging in nanoscale topography [5]) has been developed [6].

DNA-based Probes for Super-resolution Imaging: DNA-PAINT
The key innovation in DNA-PAINT is the fact, that switching of dyes between bright and dark states is facilitated by reversible DNA hybridization reactions with programmable kinetics (Fig. 1b) and thus decoupled from their photophysical properties. Targets are "labeled" with short single-stranded DNA molecules ("docking" strands). Complementary fluorescently labeled strands in solution ("imager" strands) transiently bind to these docking strands. Fluorescence readout is performed in total internal reflection (TIR), thereby in the unbound (dark) state only background fluorescence is observed. Upon binding, an imager strand is temporally immobilized on the target molecule and its fluorescence is detected. The transient binding creates an apparent "blinking" of the target, which is then used for stochastic reconstruction microscopy.

The programmability of DNA hybridization reactions allows one to independently control fluorescence bright and dark times over a wide range. Bright and dark times are varied by changing the binding strength and concentration of the imager strands. Importantly, imaging can be performed in standard DNA hybridization buffers without the use of special photoswitching reagents, such as primary thiols or reduction/oxidation systems. This makes DNA-PAINT imaging compatible with most dyes.

To demonstrate super-resolution imaging, we used the DNA origami technique to design self-assembled nanostructures with docking strands placed at prescribed positions [7]. In DNA origami, a long single-stranded DNA molecule is folded by ~200 short oligonucleotides into defined shapes and patterns.

Here, a rectangular DNA origami (90 nm x 60 nm) displays ~30 docking strands arranged in a pattern resembling the digit "0" (Fig. 1c).

The concept can also be used to image cellular targets such as proteins. Antibodies coupled with docking strands are used to label the target proteins following standard immunostaining procedures [8].

Multicolor DNA-PAINT imaging using spectrally distinct dyes is straightforward to implement, as different targets can be labeled with orthogonal DNA docking sequences. The corresponding complementary imager strands are labeled with different color dyes. The unique specificity of DNA hybridization enables multicolor imaging with no cross-talk between different colors [9].

Although using spectrally distinct fluorophores allows for the simultaneous observation of multiple targets, the total number of distinct colors is limited due to their overlapping spectra. This usually limits multiplexing in fluorescence microscopy to about 4-5 targets. Multiplexing is especially challenging in super-resolution microscopy, as photoswitching and detection requirements for multiple single-molecule dyes are often hard to implement.

DNA-based imaging probes enable a novel way of multiplexing, which we call Exchange-PAINT [8]. Here, multiplexing does not rely on multiple dyes with distinct spectral properties.

Multiplexed Super-resolution Imaging Using Exchange-PAINT
As imager strands only bind transiently to their complementary docking strands, multiplexed imaging can be performed by labeling all targets with orthogonal docking strands and then sequentially adding and removing the corresponding imager strands to the sample. This concept is illustrated in Figure 2a. After labeling all target species with orthogonal docking strands, imager strands with sequence S1 (complementary to the docking sequence S1* on the first target) are added to the sample and a super-resolution image is acquired. The transient binding of imager strands then allows us to remove them from the sample solution by a quick washing step (~1-2 min). In a second step, imager strands with sequence S2 (complementary to the docking sequence S2*) are added to the sample and image acquisition is performed in the same way as before. As imaging is performed sequentially, the same color dye can be used in each round of Exchange-PAINT. This process is repeated until all target species are imaged. Finally, pseudocolored images from all targets are superimposed and the multiplexed super-resolution image is obtained (Fig. 2b). As only a single fluorophore is used for all imaging rounds, the same high resolution for each target is maintained.

Conclusions and Outlook
Using DNA-PAINT, we demonstrated straightforward super-resolution imaging on DNA origami [6] and in fixed cells [8]. Using Exchange-PAINT, we so far showed 10-color super-resolution imaging in vitro on DNA origami and 4-color super-resolution imaging in situ in mammalian cells in two and three dimensions [8]. 3D DNA-PAINT also offers a non-destructive way of imaging 3D DNA nanostructures with high efficiency [10]. Exchange-PAINT multiplexing is only limited by the number of orthogonal DNA sequences, and not by the number of spectrally distinct dyes. This, in principle, allows us to image hundreds of different species in a single sample.

References
[1] Hell S. W.: Nat Meth 6, 24 (2009)
[2] Hell S. W. and Wichmann J.: Optics Letters 19, 780 (1994)
[3] Betzig E. et al.: Science 313, 1642 (2006)
[4] Rust M. J. et al.: Nat Meth 3, 793 (2006)
[5] Sharonov A. and Hochstrasser R. M.: PNAS 103, 18911 (2006)
[6] Jungmann R. et al.: Nano Letters 10, 4756 (2010)
[7] Rothemund P. W.: Nature 440, 297 (2006)
[8] Jungmann R. et al.: Nat Meth 11, 313 (2014)
[9] Lin C. et al.: Nat Chem 4, 832 (2012)
[10] Iinuma R. et al.: Science 344, 65 (2014)

Authors
Johannes B. Woehrstein
Prof. Dr. Peng Yin
Dr. Ralf Jungmann, Harvard Univeristy, Boston, MA, USA

 

Contact

Harvard University
3 Blackfan Circle 0
02115 Boston, Massachusetts
USA
Phone: 001/617/4321204
Telefax: 001/617/4327828

Register now!

The latest information directly via newsletter.

To prevent automated spam submissions leave this field empty.