Communication pubs.acs.org/JACS
Spectrally Resolved Super-Resolution Microscopy Unveils Multipath Reaction Pathways of Single Spiropyran Molecules Doory Kim,† Zhengyang Zhang,†,§ and Ke Xu*,†,‡ †
Department of Chemistry, University of California, Berkeley, Berkeley, California 94720, United States Division of Molecular Biophysics and Integrated Bioimaging, Lawrence Berkeley National Laboratory, Berkeley, California 94720, United States
‡
S Supporting Information *
fluorogenic reaction can help unveil rich, multipath reaction pathways. This was achieved through a wide-field singlemolecule spectroscopy and super-resolution method we recently developed, namely spectrally resolved stochastic optical reconstruction microscopy (SR-STORM).15 With SRSTORM, we previously obtained the fluorescence spectra and positions of ∼106 photoswitchable single molecules in cells.15 In this work, we generalize SR-STORM to single-molecule fluorogenic reactions by studying the isomerization of 1′,3′dihydro-1′,3′,3′-trimethyl-6-nitrospiro[2H-1-benzopyran-2,2′(2H)-indole] (6-nitro BIPS), a representative photochromic spiropyran.16,17 In its spiropyran form, 6-nitro BIPS is nonfluorescent due to perpendicularly oriented indole and benzopyran moieties (Figure 1a). Through a ring-opening reaction (Figure 1a; Process 1) activatable by thermal energy or a UV/violet light, the product merocyanine is planar and fluorescent. At room temperature, the merocyanine form is often found as two stable cis−trans isomers, commonly denoted as TTC and TTT (Figure 1a).18−24 The two isomers have different spectral characteristics, but with conventional approaches it is difficult to untangle their respective contributions in the mixture. We attempted to overcome this issue here by measuring the fluorescence emission spectra of single merocyanine molecules as they are individually generated from the ring-opening reaction. Previous studies have shown that the UV- and thermalactivated merocyanine form of 6-nitro BIPS to be good singlemolecule emitters with high fluorescence brightness and photostability,25,26 but single-molecule fluorescence spectra are not obtained due to the aforementioned difficulties. As a wide-field technique,15 SR-STORM is uniquely suitable for collecting the emission spectra of single molecules that randomly switch into the fluorescent state over large areas. 6-Nitro BIPS was spin-coated onto a poly(D-lysine)-coated glass coverslip, sealed in a chosen solvent (n-hexane, acetone, ethanol, methanol, or water) with another coverslip, and mounted between the two opposing objective lens of our SRSTORM setup (Figure 1a). Although surface-immobilization may conceivably affect reaction dynamics and spectral properties, it appears necessary for following the spectral evolution of single molecules. The application of spiropyrans is also often under similar immobilized conditions.16,17 A 560 nm laser excited the product merocyanine molecules. The application of
ABSTRACT: By recording both the images and emission spectra of thousands of single fluorescent molecules stochastically generated from the ring-opening reaction of a spiropyran, we provide mechanistic insights into its multipath reaction pathways. Through statistics of the measured single-molecule spectra, we identify two spectrally distinct isomers, presumably TTC and TTT cis−trans isomers, for the open-ring merocyanine product, and discover a strong solvent polarity-dependence for the relative population of the two isomers. From singlemolecule spectral time traces, we further examine isomerization between the two product merocyanine isomers, as well as their ring-closure reaction back to the spiropyran form.
R
ecent years have witnessed the rise of single-molecule fluorescence microscopy1−3 as a key tool to study chemical reactions at single-molecule and super-resolution levels.4−8 Building upon the notion that an initially nonfluorescent molecule may generate a fluorescent product through a fluorogenic reaction, in situ monitoring of how single-molecule fluorescence switches on and off over time has provided rich information on reaction dynamics. Moreover, through integration with high-precision localization of single fluorophores as they are generated from reactions, it has become possible to achieve reaction-based super-resolution microscopy to unveil local chemical reactivity at spatial resolutions beyond the diffraction limit of the light. However, the potentially rich dimension of spectrally resolved single-molecule fluorescence, an important factor in the development of super-resolution microscopy,9,10 remains largely unexplored for single-molecule reactions. Conventional approaches for single-molecule spectral measurement rely on locally confined, single-location illumination and detection,11−14 thus necessitating sample scanning to measure different molecules and are limited in throughput and spatial resolution. Moreover, for fluorogenic reactions, as one cannot predict a priori at what location a single, nonfluorescent (thus undetectable) reactant molecule would react to generate a fluorescent product molecule, it is difficult to follow the reaction dynamics of individual molecules through singlelocation spectral measurements. Here we demonstrate that the capability to resolve in situ the fluorescence emission spectra of single product molecules of a © 2017 American Chemical Society
Received: May 4, 2017 Published: July 3, 2017 9447
DOI: 10.1021/jacs.7b04602 J. Am. Chem. Soc. 2017, 139, 9447−9450
Communication
Journal of the American Chemical Society
By first treating the appearance of every molecule in each camera frame as a separate spectrum, we obtained ∼104 singlemolecule spectra in minutes. This allowed for a statistical examination of how different molecules behave as they are stochastically switched into the fluorescent state through ringopening (Process 1 in Figure 1a), as well as how the fluorescence spectrum of each molecule evolves over time. The concurrently measured single-molecule images in each frame (Figure 1b) further allowed for the super-resolved localization of each molecule. The resultant STORM images (Figure 1e) appeared as well-isolated nanoclusters due to the repeated localization of the same molecules across different frames. Statistics of the positional variations within each nanocluster gave standard deviations (localization uncertainties) of ∼10 nm (Figure S2), in agreement with what is typically achieved in STORM.27 In contrast, diffraction-limited conventional images, as obtained by accumulating the detected fluorescence across frames, yielded blurry images in which single molecules are not well resolved (Figure 1f). To facilitate a statistics of the emission wavelength of the ∼10 000 single-molecule spectra collected in each experiment, we calculated the spectral mean of each molecule as the intensity-weighted average of wavelength,12,15 and used this value to present the spectral position of the molecule. Remarkably, by plotting the distribution of the measured spectral mean value and photon count of every detected single merocyanine molecule in each camera frame (Figure 2a−e), we identified two major subpopulations that were comparable in brightness but distinct in emission spectrum. Spectral means of the two subpopulations centered at ∼610 and ∼630 nm, respectively (Figure 2a−e), and corresponding single-molecule spectra peaked at ∼590 and ∼635 nm (Figure 2f−h). For data recorded in n-hexane, the two subpopulations appeared comparable in total counts, but as we moved to solvents of increased polarity, the long-wavelength subpopulation diminished considerably. Increasing the starting 6-nitro BIPS concentration by 3-fold did not alter the distribution (Figure S3), suggesting that neither subpopulation was due to dimerization. The observed two subpopulations may be attributed to the aforementioned two stable isomers of merocyanine (Figure 1a). Previous work indicates that at room temperature, both the TTC and TTT isomers of merocyanines exist at significant amounts, with TTC often at a higher level.18−24 Although recent studies tend to assign the isomer with shorter absorption and emission wavelengths to TTC,22−24 the opposite assignment has been made in earlier work.21 For 6-nitro BIPS, one study suggested that the fluorescence emission of TTC and TTT isomers peaks at ∼615 and ∼630 nm, respectively.23 Our time-dependent density functional theory (TD-DFT) calculations showed emission bands at 562 and 599 nm for TTC and TTT in water, respectively. For convenience of discussion, we thus tentatively assigned the short- and long-wavelength emitting subpopulations of our single-molecule results to TTC and TTT, respectively, although alternative assignments are still possible. Based on this assignment, our results suggest that the ring-opening reaction of spiropyran leads to comparable amounts of TTC and TTT molecules in hexane, but the TTT population decreases significantly in more polar solvents (Figure 2i). The observed strong solvent polarity-dependence may be related to solvent stabilization. Our DFT calculations gave 14.4 and 14.1 D for the dipole moments of TTC and TTT isomer,
Figure 1. Spectrally resolved, super-resolution investigation of the ring-opening, ring-closure, and cis−trans isomerization reactions of single spiropyran molecules. (a) Schematic of the system. Single nonfluorescent 6-nitro BIPS spiropyran molecules undergo ringopening (Process 1) to generate fluorescent merocyanine molecules of two stable isomers, presumably TTC and TTT. The resultant merocyanine molecules may further undergo cis−trans isomerization (Processes 2 and 3) and ring-closure (Process 4) back to spiropyran. The sample is sandwiched between two coverslips, and the fluorescence spectra and super-resolved locations of the product single merocyanine molecules are simultaneously obtained from the front and back of the sample. (b,c) A small region of the concurrently acquired fluorescence images (b) and spectra (c) of three single merocyanine molecules in water, obtained in a 33 ms snapshot. (d) Representative emission brightness traces of two single molecules in water, plotted as the count of photons detected per frame (33 ms). (e,f) Comparison of STORM (e) and diffraction-limited (f) images of merocyanine molecules.
a 405 nm laser significantly promoted the otherwise thermally activated26 spontaneous ring-opening process (Figure S1), hence a means to control the density of fluorescent molecules in the sample during the experiment, similar to that in typical STORM experiments with photoswitchable dyes.15,27 Fluorescence from the product merocyanine molecules was collected from the front and back of the sample by two opposing objective lenses. Fluorescence collected by the first objective provided images of the single molecules for positional measurement (Figure 1b), and a dispersive prism was inserted into the light path of the second objective to generate spectra of the same single molecules in the wide-field (Figure 1c).15 An EM-CCD camera operated at 30 frames per second to record concurrently the single-molecule images and spectra from the same field of view of ∼40 × 40 μm. In typical experiments, ∼10 well-separated fluorescent molecules were detected in the view to avoid possible spatial overlapping of signal (Figure 1b,c for a small region of the view). Single-step switch-on and switch-off of fluorescence (Figure 1d) confirmed that the detected fluorescence was from single molecules. Single-molecule fluorescence often lasted ∼1 s (Figure 1d), and in each frame ∼2000 photons were detected for each molecule, in general agreement with previous results.25,26 9448
DOI: 10.1021/jacs.7b04602 J. Am. Chem. Soc. 2017, 139, 9447−9450
Communication
Journal of the American Chemical Society
Figure 2. Statistics of single-molecule spectra reveals two spectrally distinct isomers of the product merocyanine, and the relative populations of the two isomers depend strongly on solvent polarity. (a−e) Distribution of the measured emission spectral means and photon counts for single merocyanine molecules produced from the ring-opening reaction of spiropyran, in five solvents of increasing polarity (n-hexane, acetone, ethanol, methanol, and water). (f−h) Representative single-molecule spectra for the two isomers in n-hexane, ethanol, and water. Two spectra, each collected for one single molecule in a single camera frame (33 ms integration), are shown for each isomer. Wavelengths shorter than 570 nm were cut by emission filter. (i) Dependence of the TTT/TTC ratio based on the distribution of single-molecule spectra in (a−e), as a function of the relative polarity (normalized Dimroth−Reichardt ET scale) of the solvent.
respectively (Figure S4 for optimized structures). Previous calculations also consistently predict TTC to be more polar than TTT.28,29 The TTC isomer may thus be more stabilized in solvents of increased polarity. However, due to the nonequilibrium nature of our experiments, kinetic factors, which may also change with solvent polarity, could instead have dominated our results. Our observation of two spectrally distinct subpopulations of single molecules suggests that the two isomers do not readily convert to each other at time scales faster than our spectrum integration time (∼33 ms), which would have led to one single, intermediate spectra. Indeed, by plotting the spectral mean of individual molecules as a function of time, we often observed single molecules with fixed spectra that corresponded to either TTC or TTT (Figure 3a,b), indicating that after ring-opening from spiropyran, these merocyanine molecules are locked into one particular isomer. Occasionally we observed single-step jumps in the singlemolecule time traces of spectral mean (Figure 3c), indicative of TTC → TTT and TTT → TTC isomerization processes (Processes 2 and 3 in Figure 1a). Statistics of time traces (Figure 3d and Figure S5) indicated that in water, >∼85% of the product merocyanine molecules were locked into the TTC state, whereas in hexane, similar subpopulations (∼35%) were found in the fixed TTC and TTT states, and ∼5% underwent isomerization to either directions. The stronger TTT presence in hexane is consistent with the overall distributions of singlemolecule spectra (Figure 2a,e). Recent ultrafast spectroscopy studies have shown that a small fraction of TTC undergoes TTC → TTT isomerization through the excited state, whereas the opposite process is not observed.23,24 Our single-molecule results indicate that a small fraction of the merocyanine molecules undergoes TTC → TTT or TTT → TTC isomerization at second-time scales under our experimental
Figure 3. Monitoring the isomerization dynamics of single merocyanine molecules through spectral time traces. (a,b) Time traces of the spectral mean of two spectrally unvarying single molecules corresponding to the presumed TTC (black) and TTT (red) isomers, respectively, in water (a) and n-hexane (b). (c) Time traces showing examples of TTC → TTT isomerization (red) and TTT → TTC isomerization (black) of single molecules in n-hexane. (d) Statistics of ∼1000 single-molecule spectral time traces in water and hexane categorized into nonisomerizing TTC and TTT isomers or TTC → TTT and TTT → TTC isomerization processes. A fraction of the traces exhibited complex switching/scattering behavior that was complicated by noise, and were denoted as “unclassified” (Figure S5).
conditions, where isomerization dynamics may have been slowed down due to immobilization of the molecules.25,26 In addition to isomerization between the fluorescent TTC and TTT merocyanine states, we also occasionally observed reversible on/off fluorescence switching of single molecules. Figure 4a shows an example in which a TTC isomer switched 9449
DOI: 10.1021/jacs.7b04602 J. Am. Chem. Soc. 2017, 139, 9447−9450
Journal of the American Chemical Society
Communication
■
ACKNOWLEDGMENTS We thank K. Durkin and O. Olatunji-ojo for assistance with theoretical calculations. This work was supported by the Beckman Young Investigator Program. K.X. acknowledges additional support from Packard Fellowships for Science and Engineering and the Sloan Research Fellowship.
■
Figure 4. Reversible on/off fluorescence switching of single molecules. (a) Time trace of the spectral mean of one single molecule in water that switched on fluorescence emission for ∼2 s (red), switched off for ∼4 s, and then switched on again (blue). (b) Time trace of a single molecule in hexane that initially switched on as a TTT isomer (red), switched off for a few seconds, and then switched back on as a TTC isomer (blue). (c) Time trace of a molecule in hexane that underwent both reversible on−off switching and TTT−TTC isomerization. Insets: “+” represents the superlocalized position of the molecule in each recorded frame, colored the same way as the time traces.
off in fluorescence emission for a few seconds before switching on again. Super-resolved localization of the molecule in each imaging frame (Figure 4a−c insets) showed identical positions within localization uncertainty (∼10 nm), thus confirming that the fluorescence came from the same single molecule. The observed long switch-off time suggests that the merocyanine molecule switched to a stable dark state, possibly through ringclosure to spiropyran (Process 4 in Figure 1a). Consistent with this assignment, in the second ring-opening process, the single molecule may switch to a different merocyanine isomer, as evidenced by the emission spectrum (Figure 4b). More complex behaviors that combined reversible on/off switching and TTT-TTC isomerization were also occasionally noted (Figure 4c). Overall, we found reversible on/off switching to be rare (∼1%), an observation in agreement with previous singlemolecule studies.25 Most fluorescent merocyanine molecules thus photobleached, rather than ring-closed, at the ends of their emission. In summary, by generalizing SR-STORM to fluorogenic reactions, we unveiled multipath reaction pathways of single spiropyran molecules. The capability to track, over large sample areas, the emission spectra of single molecules that randomly switch on may also be harnessed to study other multipath reactions at the single-molecule level.
■
ASSOCIATED CONTENT
S Supporting Information *
The Supporting Information is available free of charge on the ACS Publications website at DOI: 10.1021/jacs.7b04602. Experimental details and additional data (PDF)
■
REFERENCES
(1) Xie, X. S.; Trautman, J. K. Annu. Rev. Phys. Chem. 1998, 49, 441. (2) Moerner, W. E.; Orrit, M. Science 1999, 283, 1670. (3) Ha, T. Nat. Methods 2014, 11, 1015. (4) De Cremer, G.; Sels, B. F.; De Vos, D. E.; Hofkens, J.; Roeffaers, M. B. J. Chem. Soc. Rev. 2010, 39, 4703. (5) Tachikawa, T.; Majima, T. Chem. Soc. Rev. 2010, 39, 4802. (6) Cordes, T.; Blum, S. A. Nat. Chem. 2013, 5, 993. (7) Chen, P.; Zhou, X. C.; Andoy, N. M.; Han, K. S.; Choudhary, E.; Zou, N. M.; Chen, G. Q.; Shen, H. Chem. Soc. Rev. 2014, 43, 1107. (8) Chen, T.; Dong, B.; Chen, K.; Zhao, F.; Cheng, X.; Ma, C.; Lee, S.; Zhang, P.; Kang, S. H.; Ha, J. W.; Xu, W.; Fang, N. Chem. Rev. 2017, 117, 7510. (9) Betzig, E. Opt. Lett. 1995, 20, 237. (10) van Oijen, A. M.; Kohler, J.; Schmidt, J.; Muller, M.; Brakenhoff, G. J. Chem. Phys. Lett. 1998, 292, 183. (11) Macklin, J. J.; Trautman, J. K.; Harris, T. D.; Brus, L. E. Science 1996, 272, 255. (12) Lu, H. P.; Xie, X. S. Nature 1997, 385, 143. (13) Trautman, J. K.; Macklin, J. J.; Brus, L. E.; Betzig, E. Nature 1994, 369, 40. (14) Moerner, W. E.; Fromm, D. P. Rev. Sci. Instrum. 2003, 74, 3597. (15) Zhang, Z. Y.; Kenny, S. J.; Hauser, M.; Li, W.; Xu, K. Nat. Methods 2015, 12, 935. (16) Berkovic, G.; Krongauz, V.; Weiss, V. Chem. Rev. 2000, 100, 1741. (17) Klajn, R. Chem. Soc. Rev. 2014, 43, 148. (18) Hobley, J.; Malatesta, V.; Millini, R.; Montanari, L.; Parker, W. O. N. Phys. Chem. Chem. Phys. 1999, 1, 3259. (19) Hobley, J.; Malatesta, V. Phys. Chem. Chem. Phys. 2000, 2, 57. (20) Hobley, J.; Pfeifer-Fukumura, U.; Bletz, M.; Asahi, T.; Masuhara, H.; Fukumura, H. J. Phys. Chem. A 2002, 106, 2265. (21) Wohl, C. J.; Kuciauskas, D. J. Phys. Chem. B 2005, 109, 22186. (22) Buback, J.; Nuernberger, P.; Kullmann, M.; Langhojer, F.; Schmidt, R.; Wurthner, F.; Brixner, T. J. Phys. Chem. A 2011, 115, 3924. (23) Ruetzel, S.; Diekmann, M.; Nuernberger, P.; Walter, C.; Engels, B.; Brixner, T. J. Chem. Phys. 2014, 140, 224310. (24) Nuernberger, P.; Ruetzel, S.; Brixner, T. Angew. Chem., Int. Ed. 2015, 54, 11368. (25) Hu, D. H.; Tian, Z. Y.; Wu, W. W.; Wan, W.; Li, A. D. Q. J. Am. Chem. Soc. 2008, 130, 15279. (26) Simoncelli, S.; Roberti, M. J.; Araoz, B.; Bossi, M. L.; Aramendia, P. F. J. Am. Chem. Soc. 2014, 136, 6878. (27) Rust, M. J.; Bates, M.; Zhuang, X. W. Nat. Methods 2006, 3, 793. (28) Balasubramanian, G.; Schulte, J.; Muller-Plathe, F.; Bohm, M. C. Chem. Phys. Lett. 2012, 554, 60. (29) Amos, R. D.; Kobayashi, R. Mol. Phys. 2015, 113, 1674.
AUTHOR INFORMATION
Corresponding Author
*
[email protected] ORCID
Ke Xu: 0000-0002-2788-194X Present Address §
Division of Chemistry and Biological Chemistry, School of Physical and Mathematical Sciences, Nanyang Technological University, Singapore 637371 Notes
The authors declare no competing financial interest. 9450
DOI: 10.1021/jacs.7b04602 J. Am. Chem. Soc. 2017, 139, 9447−9450
