Correlative Single-Molecule FRET and DNA-PAINT Imaging - Nano

Jun 26, 2018 - DNA-PAINT is an optical super-resolution microscopy method that can visualize nanoscale protein arrangements and provide spectrally ...
1 downloads 0 Views 2MB Size
Download PDF
Letter Cite This: Nano Lett. 2018, 18, 4626−4630

pubs.acs.org/NanoLett

Correlative Single-Molecule FRET and DNA-PAINT Imaging Nina S. Deußner-Helfmann,† Alexander Auer,‡,§ Maximilian T. Strauss,‡,§ Sebastian Malkusch,† Marina S. Dietz,† Hans-Dieter Barth,† Ralf Jungmann,*,‡,§ and Mike Heilemann*,† †

Single Molecule Biophysics, Institute of Physical and Theoretical Chemistry, Goethe-University Frankfurt, Max-von-Laue-Strasse 7, 60438 Frankfurt, Germany ‡ Department of Physics and Center for Nanoscience, Ludwig Maximilian University, 80539 Munich, Germany § Max Planck Institute of Biochemistry, 82152 Martinsried, Germany Downloaded via UNIV OF SUNDERLAND on November 2, 2018 at 09:36:09 (UTC). See https://pubs.acs.org/sharingguidelines for options on how to legitimately share published articles.

S Supporting Information *

ABSTRACT: DNA-PAINT is an optical super-resolution microscopy method that can visualize nanoscale protein arrangements and provide spectrally unlimited multiplexing capabilities. However, current multiplexing implementations based on, for example, DNA exchange (such as ExchangePAINT) achieves multitarget detection by sequential imaging, limiting throughput. Here, we combine DNA-PAINT with single-molecule FRET and use the FRET efficiency as parameter for multiplexed imaging with high specificity. We demonstrate correlated single-molecule FRET and superresolution on DNA origami structures, which are equipped with binding sequences that are targeted by pairs of dye-labeled oligonucleotides generating the FRET signal. We futher extract FRET values from single binding sites that are spaced just ∼55 nm apart, demonstrating super-resolution FRET imaging. This combination of FRET and DNA-PAINT allows for multiplexed super-resolution imaging with low background and opens the door for accurate distance readout in the 1−10 nm range. KEYWORDS: Single-molecule FRET, DNA-PAINT, super-resolution microscopy, DNA origami, multiplexing

S

low imager strand concentrations, which in turns leads to rather long acquisition times to achieve sufficient target sampling. To minimize background and increase imaging speed, DNAPAINT has recently been combined with single molecule Förster resonance energy transfer (smFRET).8,9 Instead of one imager strand, FRET-PAINT uses two different imager strands where one carries a donor and the other an acceptor dye. Upon simultaneous binding of both donor and acceptor strands, FRET occurs, and a fluorescence signal can be observed in the acceptor channel. As the background signal originates from the excitation of the donor fluorophore and thus is not present in the acceptor channel, background signal can be effectively minimized. This allows the use of higher imager strand concentrations, resulting in faster image acquisition. Another advantage is the reduction of an unspecified signal because two probes have to bind simultaneously to the same target. However, due to the specific distance dependency of smFRET FRET-PAINT is ideally suited to identify different targets by FRET efficiencies. This information has not been used so far. Here, we combine the super-resolution capabilities of DNA-PAINT to resolve nanoscale targets with smFRET information on fluorophore distances. For the first time to our knowledge, a specific design of imager strands enables the

ingle-molecule localization microscopy (SMLM) allows the visualization of structures below the optical diffraction limit. Conventional SMLM relies on the stochastic photoswitching of fluorophores between a fluorescent and a nonfluorescent state, which enables time-resolved separation of single dye molecules.1 For example, this can be achieved by photoactivation or photoconversion of fluorescent proteins, as in photoactivated-localization microscopy (PALM),2 or by photoswitchable organic dyes, as in (direct) stochastic optical reconstruction microscopy ((d)STORM).3,4 Another option is to use fluorophores that transiently bind to their target and otherwise freely diffuse in an imaging buffer, as in point accumulation in nanoscale topography (PAINT).5 DNAPAINT is an extension of PAINT that uses short, dye-labeled DNA strands (imager strands) that sequence-specifically and transiently bind to a complementary target DNA molecule (docking strand). While the programmable DNA sequence allows precise tuning of on- and off-rates, the reversible binding of imagers to docking strands minimizes the impact of photobleaching as imager strands are constantly replenished from solution. DNA-PAINT furthermore can be performed with virtually any dye independent of its photophysical behavior and has spectrally unlimited multiplexing capabilities.6,7 However, a major limitation of DNA-PAINT is a high background signal generated by freely diffusing imager strands in the buffer. This typically limits DNA-PAINT to selective plane illumination and detection (such as total internal reflection fluorescence (TIRF) microscopy) with relatively © 2018 American Chemical Society

Received: May 30, 2018 Revised: June 22, 2018 Published: June 26, 2018 4626

DOI: 10.1021/acs.nanolett.8b02185 Nano Lett. 2018, 18, 4626−4630

Letter

Nano Letters

Figure 1. Principle of DNA-PAINT and FRET-PAINT. (a) Scheme of designed DNA origami. Every hexagon represent a single staple strand. The purple highlighted staple strands carry DNA extensions to function as docking strands for DNA-PAINT or FRET-PAINT respectively. The distances between the docking strands are ∼55 nm. (b) In DNA-PAINT, short (9 nt) Cy3B-labeled oligonucleotide strands (donor imager strands) reversibly bind to complementary target DNA strands (docking strands). Super-resolved fluorescence images are reconstructed by localizing single fluorescence emitters in a time-lapsed fashion. (c) Pseudodiffraction-limited fluorescence image (left) of DNA origami labeled with DNA-PAINT docking strands obtained by calculating a standard deviation image of a movie with 15000 frames (scale bar: 2 μm). Single super-resolved DNA origami are shown for the regions of interest (right, scale bar: 50 nm). (d) In FRET-PAINT donor (9 nt, green) and acceptor (10 nt, red) labeled DNA strands transiently bind to the complementary target strand. For the observation of a FRET signal (i.e., acceptor emission upon donor excitation) both strands need to be bound at the target at the same time. (e) Pseudodiffraction-limited fluorescence image of donor (green) and acceptor (red) emission in a sample containing high FRET probes (left, scale bar: 2 μm). After localizing single dyes, a super-resolution image is generated and the designed triangle pattern can be resolved in both channels (right, scale bar: 50 nm).

order to produce an acceptor signal. Thus, a false positive signal from single imager strands binding to nontargets is minimized. The acceptor strand is functionalized with a fluorophore at the 3′- end (ATTO 647N, Figure 2a,b). Two donor strands carrying a fluorophore (Cy3B) either at the 3′end (for high FRET) or at the 5′-end (for low FRET) were designed (Figure 2a,b). In our case, the docking strands were prepared with a binding site for a 10 nt imager strand carrying the acceptor fluorophore (acceptor strand), and a 9 nt binding region for the imager strand carrying the donor fluorophore (donor strand). The sequence of docking strands contains no spacer between the donor and acceptor binding region leading to distances of donor and acceptor of 3.4 nm with the pair of imager strands generating a high FRET signal and a distance of 6.4 nm with imager strands generating a low FRET signal. The expected FRET efficiencies calculated with a Förster radius R0 = 6.2 nm10 are EHF = 0.97 and ELF = 0.45 respectively. However, the FRET efficiencies we report in this work represent relative and uncorrected values; determining absolute distances is possible by extracting correction factors from the single-molecule FRET data. The design of the oligonucleotide imager strands was adapted from previous work with off-rates koff of 0.1 s−1 and 1 s−1 for donor and acceptor strand, respectively.8 Because of this design, the acceptor binds longer to the docking strand than the donor to increase the probability of simultaneous binding. Transient binding of donor and acceptor strands retains DNA-PAINT’s immunity to photobleaching. Correlative FRET-PAINT experiments were carried out using 20 nM acceptor and 10 nM donor strand concentration,

detection of different FRET efficiencies which we successfully calculated from super-resolved DNA-PAINT images. Thereby, we could assign FRET populations to specific target strands. Thus, correlative FRET-PAINT extends the spectrum of applications of conventional DNA-PAINT, introducing FRET as a reporter for multiplexed imaging. We showed this concept on DNA origami with binding sites that are targeted by two imager strands for FRET read-out. Results and Discussion. We designed a rectangular DNA origami structure equipped with three docking strands as target sites to establish combined smFRET and DNA-PAINT imaging (Figure 1a). The distance between the docking strands was ∼55 nm, which on the one hand is below the diffraction limit of light and needs super-resolution microscopy to be resolved and on the other hand is large enough to prevent FRET between them. First, we tested our DNA origami with conventional DNAPAINT, where fluorescently labeled imager strands transiently bind to their respective target sites (Figure 1b). The reconstructed super-resolution image shows that the individual binding sites in single DNA origami are resolved (Figure 1c; note that the intensity reflects the number of localizations and thus binding events at a particular target site). In the next step, we characterized the DNA origami carrying three docking sites designed to bind a specific pair of imager strands that gives rise to a high FRET signal. In FRET-PAINT, two different imager strands functionalized with a donor and an acceptor dye, which are reversibly hybridized to the target strands, are utilized (Figure 1d,e). Both donor and acceptor imager strands have to simultaneously bind to the target in 4627

DOI: 10.1021/acs.nanolett.8b02185 Nano Lett. 2018, 18, 4626−4630

Letter

Nano Letters

Figure 2. FRET-PAINT allows identification of FRET efficiencies E at the nanoscale. (a,b) Experimental design of oligonucleotide strands for FRET-PAINT: an acceptor fluorophore-labeled imager strand (ATTO 647N, 10 nt) (red) is combined with a second imager strand (9 nt) (green) carrying a donor fluorophore (Cy3B) either at the 5′ (high FRET, a) or 3′ end (low FRET, b). (c,d) Correlative single-molecule FRET and DNAPAINT imaging on two different DNA origami that are decorated each with three target strands binding one pair of imager strands for (c) high FRET and (d) low FRET. The individual panels show the super-resolution image of the donor channel (center) and the acceptor channel (right) (note that the intensities in the images reflect the number of binding events) (scale bars: 50 nm). (e,f) The FRET histograms are calculated by analyzing the donor and acceptor intensity at each binding event of a high (e) and a low FRET sample (f). (e) A donor only population (EDO = 0.05 ± 0.05, SD) as well as a high FRET population (EHF = 0.78 ± 0.13, SD) are identifiable. The number of independent measurements Nmeasurements were 16, the number of analyzed binding sites Nbinding sites were 1002. (f) A donor only population (EDO = 0.05) as well as a low FRET population (ELF = 0.25 ± 0.13, SD) can be distinguished (Nmeasurements = 17, Nbinding sites = 1047). (g,h) FRET-PAINT image with color-coded FRET efficiencies for (g) high FRET and (h) low FRET. Extracted FRET efficiencies from super-resolved individual binding sites show the expected populations: (g) EHF = 0.81 ± 0.09 (SD) and (h) ELF = 0.20 ± 0.02 (SD) (scale bar: 50 nm).

respectively. The fluorescence emission was detected with a dual-channel, single-molecule sensitive microscope employing shadowless TIRF illumination with a rotating laser beam.11 The reconstructed super-resolution images show that individual binding sites can be resolved in both spectral channels (Figure 2c). On average, we detected ∼1300 photons per frame for the donor fluorophore, and ∼2300 photons per frame for the acceptor fluorophore. An analysis of the experimental localization precision using a nearest neighbor analysis12 provided a value of 11.3 nm for the donor and 15.1 nm for the acceptor channel. From the super-resolved images, we were able to calculate FRET efficiencies (E) for each single binding site by analyzing the donor and acceptor intensity at each binding event (Figures S-1). A FRET histogram for a total of 1002 binding sites from 16 measurements was generated and the expected

high FRET population with EHF = 0.78 is obvious (Figure 2e). Donor leakage which is caused by bleed through of donor fluorescence into the acceptor channel results in a second population with EDO = 0.05. By calculating the FRET efficiencies of single binding sites and fitting these to a mixed model of FRET populations (donor only and high FRET or low FRET respectively), we could further explicitly identify docking sites as high FRET targets (Figure 2g). Next, we designed a DNA origami that carried three docking sites for the donor strand that in combination with the acceptor strand generates a low FRET efficiency. Using identical imaging conditions, we could clearly resolve individual binding sites in single DNA origami in both spectral channels (Figure 2d). On average, we detected ∼3300 photons per frame for the donor fluorophore, and ∼1450 photons per frame for the acceptor fluorophore. We determined an 4628

DOI: 10.1021/acs.nanolett.8b02185 Nano Lett. 2018, 18, 4626−4630

Letter

Nano Letters experimental localization precision of 9.3 nm in the donor and 26.8 nm in the acceptor channel. The much lower localization precision in the acceptor channel is explained by the lower photon number emitted by the acceptor in low FRET samples due to less energy transfer. In the FRET histogram (N = 1047 binding sites from 17 measurements), we observed two populations, a first one with ELF = 0.25 which represents the low FRET pair and a second one with EDO = 0.05 caused by donor leakage (Figure 2f). We noticed that not all origami carried the designed three binding sites in the donor and acceptor channel (Figure S-2). We quantified the number of DNA origami in which we detected two or three donor binding sites. Using a binomial distribution, we determined a detection efficiency of 64% for a single docking strand (which translates into 26% probability to detect all three binding sites in an origami) (Figure S-3a). This might to some extent be explained by docking strands which are incompletely incorporated or incorporated but inaccessible for imager strands.13 For the acceptor binding sites, we determined a detection efficiency of 54% (Figure S-3b). This somewhat smaller number results from the necessity that both the donor and the acceptor imager strand must be bound to the target. We found that 25% of DNA origami that showed three donor fluorophores also showed three acceptor fluorophores. In order to elucidate the multiplexing capacity of FRETPAINT, we analyzed a mixture of origami carrying explicitly either high or low FRET docking strands in a distinct mixture of 1:1. We were able to separate both populations in the FRET histogram. Furthermore, we could assign single DNA origami to high or low FRET species (Figure S-4). To prove that we can even discriminate single DNA docking sites within the dimension of a DNA origami structure spaced at a subdiffraction distance of ∼55 nm, we designed a DNA structure with one docking strand binding a pair of imager strands with high FRET, and two docking strands binding a pair of imager strands generating low FRET. The resulting amount of docking sites for donor strands with high FRET is one-third compared to the origami described above and twothirds for donor strands with low FRET. Because of this, the single imager strands were added at concentrations of 20 nM acceptor strand, 3.3 nM donor strand for high FRET, and 6.6 nM donor strand for low FRET. The individual binding sites in single DNA origami were resolvable with an average localization precision of 14.5 nm in the donor and 23.1 nm in the acceptor channel (Figure 3a). We could clearly distinguish the two expected populations in the FRET histogram (N = 1038 binding sites from 8 measurements) with ELF = 0.21 and EHF = 0.79 (Figure 3b). These values correspond to the previously characterized FRET pairs. The donor only population occurred at EDO = 0.04. By calculating the mean FRET efficiencies for individual binding sites, we are able to assign each docking site to one of the two FRET populations. This enables for clear identification of the target sequence by the FRET efficiency (Figure 3c). Because of the conserved distance of double-stranded DNA, DNA-PAINT in combination with smFRET allows for the precise adjustment of the FRET efficiency by sequence design. In conclusion, FRET-PAINT enables nanoscale imaging with DNA-PAINT and the determination of FRET values from individual targets that are spaced at distances shorter than the diffraction limit. With that, FRET adds an alternative route for multiplexed imaging with DNA-PAINT, while minimizing background caused by unspecific binding. Moreover, FRET-

Figure 3. FRET-PAINT for multiplexed detection of different target DNA strands. (a) Scheme of the DNA origami that is decorated with two target strands binding the pair of imager strands for low FRET (turquoise) and one target strand binding high FRET imager strands (purple) (left). Exemplary super-resolved images of donor (center) and acceptor (right) are shown for one DNA origami. The individual panels show the single-molecule localization image of the donor channel (left), acceptor channel (right) (note that the intensities in the images reflect the number of binding events) (b) The FRET histogram shows a donor only (EDO = 0.04), a low FRET population (ELF = 0.21), and a high FRET population (EHF = 0.79) (Nmeasurements = 8, Nbinding sites = 1038). FRET-PAINT image with color-coded FRET efficiencies (right). Numbers indicate the averaged FRET efficiencies as extracted from individual binding sites (scale bars: 50nm).

PAINT enables the measurement of distances in the range of 1−10 nm, hereby extending the spatial resolution of DNAPAINT to the near-molecular level. Correlated single-molecule FRET and DNA-PAINT imaging can be integrated into various and well-established applications of DNA-PAINT imaging, for example, the visualization of nanoscale protein arrangements or mRNA patterns.14,15 We further envision applications that allow monitoring protein conformations while at the same time detecting protein localization with subdiffraction resolution.

■

ASSOCIATED CONTENT

S Supporting Information *

The Supporting Information is available free of charge on the ACS Publications website at DOI: 10.1021/acs.nanolett.8b02185. Supplementary figures showing the intensity time traces of a single high FRET pair, examples of FRET-PAINT images with low and high FRET as well as of a mixture of low and high FRET DNA origami; all materials and methods including the DNA origami sequences (PDF)

■

AUTHOR INFORMATION

Corresponding Authors

*E-mail: [email protected]. *E-mail: [email protected]. ORCID

Marina S. Dietz: 0000-0003-0504-1824 Ralf Jungmann: 0000-0003-4607-3312 Mike Heilemann: 0000-0002-9821-3578 4629

DOI: 10.1021/acs.nanolett.8b02185 Nano Lett. 2018, 18, 4626−4630

Letter

Nano Letters Author Contributions

M.H. and R.J. designed the study. S.M., M.D., and H.-D.B. built the TIRF-FRET microscope with rotating excitation. N.S.D.-H. prepared samples and acquired data. A.A. designed DNA origami structures, and M.T.S. wrote a custom software for data analysis. N.S.D.-H., A.A., and M.T.S. analyzed the data. All authors discussed the data and wrote the manuscript. Notes

The authors declare no competing financial interest.

■

ACKNOWLEDGMENTS M.H. and N.S.D.-H. acknowledge support from the DFG (SFB 902). A.A. acknowledges support from the DFG through the Graduate School of Quantitative Biosciences Munich (QBM). M.T.S. acknowledges support from the International Max Planck Research School for Molecular and Cellular Life Sciences (IMPRS-LS).

■

REFERENCES

(1) Sauer, M.; Heilemann, M. Chem. Rev. 2017, 117 (11), 7478− 7509. (2) Betzig, E.; Patterson, G. H.; Sougrat, R.; Lindwasser, O. W.; Olenych, S.; Bonifacino, J. S.; Davidson, M. W.; Lippincott-Schwartz, J.; Hess, H. F. Science 2006, 313 (5793), 1642−5. (3) Heilemann, M.; van de Linde, S.; Schuttpelz, M.; Kasper, R.; Seefeldt, B.; Mukherjee, A.; Tinnefeld, P.; Sauer, M. Angew. Chem., Int. Ed. 2008, 47 (33), 6172−6. (4) Rust, M. J.; Bates, M.; Zhuang, X. Nat. Methods 2006, 3 (10), 793−5. (5) Sharonov, A.; Hochstrasser, R. M. Proc. Natl. Acad. Sci. U. S. A. 2006, 103 (50), 18911−6. (6) Jungmann, R.; Steinhauer, C.; Scheible, M.; Kuzyk, A.; Tinnefeld, P.; Simmel, F. C. Nano Lett. 2010, 10 (11), 4756−61. (7) Schnitzbauer, J.; Strauss, M. T.; Schlichthaerle, T.; Schueder, F.; Jungmann, R. Nat. Protoc. 2017, 12 (6), 1198−1228. (8) Auer, A.; Strauss, M. T.; Schlichthaerle, T.; Jungmann, R. Nano Lett. 2017, 17 (10), 6428−6434. (9) Lee, J.; Park, S.; Kang, W.; Hohng, S. Mol. Brain 2017, 10 (1), 63. (10) Holden, S. J.; Uphoff, S.; Hohlbein, J.; Yadin, D.; Le Reste, L.; Britton, O. J.; Kapanidis, A. N. Biophys. J. 2010, 99 (9), 3102−11. (11) Ellefsen, K. L.; Dynes, J. L.; Parker, I. PLoS One 2015, 10 (8), e0136055. (12) Endesfelder, U.; Malkusch, S.; Fricke, F.; Heilemann, M. Histochem. Cell Biol. 2014, 141 (6), 629−38. (13) Strauss, M. T.; Schueder, F.; Haas, D.; Nickels, P. C.; Jungmann, R. Nat. Commun. 2018, 9 (1), 1600. (14) Jungmann, R.; Avendano, M. S.; Dai, M.; Woehrstein, J. B.; Agasti, S. S.; Feiger, Z.; Rodal, A.; Yin, P. Nat. Methods 2016, 13 (5), 439−42. (15) Jungmann, R.; Avendano, M. S.; Woehrstein, J. B.; Dai, M.; Shih, W. M.; Yin, P. Nat. Methods 2014, 11 (3), 313−8.

4630

DOI: 10.1021/acs.nanolett.8b02185 Nano Lett. 2018, 18, 4626−4630