Article pubs.acs.org/JACS
Light-Induced Fluorescence Modulation of Quantum Dot-Crystal Violet Conjugates: Stochastic Off−On−Off Cycles for Multicolor Patterning and Super-Resolution Sungwook Jung,†,# Joonhyuck Park,‡,# Jiwon Bang,∥ Jae-Yeol Kim,§ Cheolhee Kim,§ Yongmoon Jeon,§ Seung Hwan Lee,† Ho Jin,‡ Sukyung Choi,‡ Bomi Kim,‡ Woo Jin Lee,‡ Chan-Gi Pack,⊥ Jong-Bong Lee,†,§ Nam Ki Lee,†,§,¶ and Sungjee Kim*,†,‡ †
School of Interdisciplinary Bioscience and Bioengineering, ‡Department of Chemistry, §Department of Physics, Pohang University of Science and Technology (POSTECH), 77 Cheongam-ro, Namgu, Pohang 37673, South Korea ∥ Korea Institute of Ceramic Engineering and Technology, 101 Soho-ro, Jinju-si, Gyeongnam 52851, South Korea ⊥ Department of Convergence Medicine, University of Ulsan College of Medicine & Asan Institute for Life Sciences, Asan Medical Center, 88 Olympic-ro 43-gil, Songpa-gu, Seoul 05505, South Korea ¶ Department of Chemistry, Seoul National University, Seoul 08826, South Korea S Supporting Information *
ABSTRACT: Photoswitching or modulation of quantum dots (QDs) can be promising for many fields that include display, memory, and super-resolution imaging. However, such modulations have mostly relied on photomodulations of conjugated molecules in QD vicinity, which typically require high power of high energy photons at UV. We report a visible light-induced facile modulation route for QD-dye conjugates. QD crystal violets conjugates (QD-CVs) were prepared and the crystal violet (CV) molecules on QD quenched the fluorescence efficiently. The fluorescence of QD-CVs showed a single cycle of emission burst as they go through three stages of (i) initially quenched “off” to (ii) photoactivated “on” as the result of chemical change of CVs induced by photoelectrons from QD and (iii) back to photodarkened “off” by radical-associated reactions. Multicolor on-demand photopatterning was demonstrated using QD-CV solid films. QD-CVs were introduced into cells, and excitation with visible light yielded photomodulation from “off” to “on” and “off” by nearly ten fold. Individual photoluminescence dynamics of QD-CVs was investigated using fluorescence correlation spectroscopy and single QD emission analysis, which revealed temporally stochastic photoactivations and photodarkenings. Exploiting the stochastic fluorescence burst of QD-CVs, simultaneous multicolor super-resolution localizations were demonstrated.
■
INTRODUCTION Quantum dots (QDs) are unique nanofluorophores that show unparalleled wavelength tunability and suitability for multiplexing.1 An ability to modulate the emission state of QD or QD conjugate would have been exploited in many applications including memory,2,3 display,4,5 diagnosis,6 sensing,7,8 and imaging.9−11 Unlike small molecules that can be photochromic, it is challenging to make QDs interplay with external stimuli such as electromagnetic wave. For the modulation of QD photoluminescence (PL), interaction between QD and conjugated molecules in the vicinity is typically used. When QDs are conjugated with molecules that bear a photoresponsive or photochromic unit, illumination can induce structural change in the conjugated molecules which in turn modulates the QD PL.12,13 For example, QDs tethered with polyviologen4 or a photocleavable ortho-nitrobenzyl group10,14 showed photoactivation. Photochromic spiropyrans15 or diheteroarylethenes16 were conjugated to QDs so that energy transfers from QD to photochromic groups of different states modulated the QD PL. © 2017 American Chemical Society
Dopamine-conjugated QDs have been used as a cellular pH sensor by exploiting the pH-dependent equilibrium between dopamine and the reduced quinone form which quenches the QD PL by electron transfer. 9 In our previous study, acridinedione-conjugated QDs can be photomodulated by additional UV excitations that induce the electron transfers to QDs and quench their PL.17 Interconversion of QD-conjugated molecules typically requires relatively long irradiation of strong UV. In addition, QDs usually have orders of magnitude stronger UV absorption than the conjugated molecules, so the efficiency of such modulations is low. Dopant-level excitation can be used for QD modulation, but this strategy is limited to doped QDs and typically requires high-power excitation.18 Photomodulations of QDs have mostly relied on a process initialized by photoabsorption by conjugated molecules near the QD, and this process is inefficient and requires high-energy photons. Herein, Received: March 14, 2017 Published: May 11, 2017 7603
DOI: 10.1021/jacs.7b02530 J. Am. Chem. Soc. 2017, 139, 7603−7615
Article
Journal of the American Chemical Society
Figure 1. QD PL quenching by the CV conjugation and photoactivation and photodarkening of QD-CVs. (a) Absorption (black) and emission (red) spectra of the QD, and absorption spectrum (violet) of CV. (b) PL spectra of QD-CV samples with different CV-to-QD ratios: 3.5 (brown), 7 (red), 10.5 (pink), 14.0 (orange), 17.5 (yellow), 21 (green), 28 (green blue), and 35 (blue) along with the PL spectrum of the unconjugated QD sample (black). (c) Absorption and emission spectra of the QD (black) and QD-3.5CV (brown). (d,e) Photoactivation time evolution of PL (d) and absorption (e) spectra for QD-3.5CV sample under green irradiation (532 nm, 0.3 mW/cm2) at 0 (brown), 30 (red), 60 (green), and 120 (blue) min. (f) Photodarkening time evolution of PL spectra for QD-3.5CV sample under green irradiation (532 nm, 0.3 mW/cm2) at 120 (blue), 300 (green), 390 (red), and 480 (brown) min. (g) A scheme for PL modulation of a QD-CV. A CV is shown electrostatically conjugated to the surface molecule of the QD. PL of QD-CV is quenched by the electron transfer from QD to CV. CV is photodegraded to BDBP, which activates the PL of QD. Continued light irradiation induces radical-associated photodarkening. For the CV and BDBP, some hydrogen atoms are number-labeled for NMR assignments. Left: molecular structure of sulfonate-terminated ligand, 2-(5-(1,2-dithiolan-3-l)pentanamido)ethanesulfonate. PBS buffer (10 mM, pH7.4) by using two-phase exchange.19 The assynthesized QDs were ligand-exchanged with sulfonate-teminated surface ligands. Excess amounts (>106 time the number of QDs) of surface ligands (oxidized form) were dissolved in PBS buffer (10 mM, pH 7.4). Sodium borohydride (NaBH4) to QD (NaBH4-to-QD ratio = 2) was added to the ligand solution and vigorously stirred for at least 30 min at RT. The NaBH4 cleaves the disulfide bond which changes to dimercapto anchoring group. The QD solution (in CHCl3) was added to the reduced ligand solution and further stirred for 12 h at RT. The QDs were transferred from the organic layer to PBS buffer layer. To remove excess free surface ligands, the QD solution was dialyzed three times using an Amicon centrifugal filter (MW cutoff: 50 kDa). Photopatterning Measurement of QD-CV films. QD-CV films were placed under a fluorescence microscope (Axioplan2, Zeiss) and the PL of QD-CV films was monitored under visible-light irradiation. The light of a 100-W mercury lamp (HBO100, Zeiss) was passed through a 470/40 nm (Zeiss) or 610/30 nm (Chroma) bandpass filter and a 10× objective lens (Zeiss). Exposure times were 20 ms and all images were captured under identical camera conditions. Images were collected and analyzed using AxioVision (Zeiss). On the basis of mercury lamp irradiance power and focal area, we calculated the irradiation power density to be 11 mW/cm2 at 470 nm and 30 mW/cm2 at 610 nm.
we report a QD photomodulation route initiated by photoabsorption by QDs. Because the process begins with light absorption by the QD, the modulation wavelength can be in the visible or infrared range, depending on the QD band gap. QD crystal violets conjugates (QD-CVs) were prepared and the crystal violet (CV) molecules on QD quenched the fluorescence efficiently. The photoexcited electrons in QD are designed to transfer to CVs that are conjugated to its surface. Upon excitation, photoexcited electrons in QDs transfer to CVs, and the electron-doped CVs degrade to a nonquenching species which in turn activates the QD PL. Interestingly, continued photoexcitations quenched the QD PL as a result of radical associated process. The PL of QD-CVs showed a single cycle of emission burst while passing through three stages from initial quenched “off” to photoactivated “on” and back to photodarkened “off”.
■
EXPERIMENTAL SECTION
Surface Modification of QDs with Sulfonate-Terminated Ligands. The as-synthesized QDs were transferred from CHCl3 to 7604
DOI: 10.1021/jacs.7b02530 J. Am. Chem. Soc. 2017, 139, 7603−7615
Article
Journal of the American Chemical Society PL Measurement of Cellular QDs or QD-CVs. Human cervical carcinoma HeLa cells were incubated in RPMI 1640 (HyClone) supplemented with 10% fetal bovine serum (FBS, GIBCO) and 1% penicillinstreptomycin (PS). Cells were grown in coverslip-bottom 24well plates at a density of 1 × 105 cells/well at 37 °C under 5% CO2. After 24 h, cells were incubated with 50 nM QDs or QD-CVs (CV-toQD ratio = 35) for 2 h. Cells on coverslips were fixed using 4% formaldehyde for 20 min and rinsed three times with PBS buffer. The fixed cells were mounted on slide glass by using aqueous mounting medium with antifading agent (Biomeda). The fixed cells on the glass were placed under the fluorescence microscope (Axioplan2, Zeiss) and the PL of QD in the cells was monitored under visible light illumination. The light of a 100-W mercury lamp (HBO100, Zeiss) was passed through a 470/40 nm bandpass filter (Zeiss) and 40× objective lens (0.3 NA, Zeiss). Exposure times were 100 ms and all images were captured under identical camera conditions. Images were collected and analyzed using AxioVision (Zeiss). On the basis of mercury lamp irradiance power at 470 nm and focal area, the irradiation power density was 11 mW/cm2. Total Internal Reflection Fluorescence (TIRF) Microscope Setup and Localization of Single QD-CVs. Image time series of 4000−12 000 frames of QDs or QD-CVs with 100 ms exposure time were collected (Scheme S1) on a home-built prism-type TIRF microscope with an electron-multiplying charge-coupled device (EMCCD) camera (iXon3 897, Andor). A 405 nm laser (MLD 405 nm cw diode laser, Cobolt) was focused into a FS Pellin-Broca prism (EKSMA optics) placed on top of a quartz slide with a thin layer of immersion oil. The QDs or QD-CVs were excited by the laser. We adjusted the laser power (0.5 W/cm2 to 3.3 W/cm2) by using a circular neutral-density filter (NDC-50C-4M, Thorlabs) to investigate excitation power-dependent PL dynamics. The PL from QDs or QD-CVs was collected using a water-immersion objective lens (UPlanSApo 60× , NA1.2 objective, Olympus) and enlarged by a 1.6× magnifier. Final magnification was 96×. The emission beam was split to green and red emission by a 550 nm dichroic mirror (Photometericsoptics). Emissions with wavelength 550 nm go through the emission filter centered at 625 nm with a bandwidth of 30 nm (245617, Chroma). The filtered PL beams were projected side-by-side onto the EMCCD camera and imaged using a custom program written in C++. The spectra detected by the camera were stored and analyzed using a personal computer. All data were obtained at room temperature.
and the positively charged CVs. The conjugated CVs efficiently quenched the PL of QD. As the CV-to-QD conjugation ratio was varied as 3.5, 7, 10.5, 14, 17.5, 21, 28, and 35, the PL decreased to 45, 33, 25, 19, 15, 13, 10, and 7% of the initial (Figure 1b, Figure S2a). CV may quench the QD PL by both energy transfer and electron transfer. Energy diagrams indicate that CV can readily accept electrons from QD (Figure S2b); this transfer is believed to be a major quenching mechanism. As a control experiment, CdTeSe-alloyed QDs were prepared (Supporting Information); they emit at infrared wavelengths (peak at 776 nm), and their emission spectrum has no overlap with the absorption of CV (Figure S3a) and thus cannot quench the QD PL by energy transfer. The CV conjugation was repeated for the CdTeSe QDs, and the PL quenching efficiency was comparable to the CdSe/ CdS/ZnS QD case (Figure S3b). Stern−Volmer plots showed similar quenching slopes for the two kinds of QDs (KSV = 1.1 × 106 M−1 for CdSe/CdS/ZnS QD and KSV = 1.2 × 106 M−1 for CdTeSe QD). CVs can quench QD PL efficiently via electron transfer. Semiconductors such as TiO223and ZnO24 are widely used as photocatalysts in wastewater purification, especially for decomposing dye pollutants. Triphenylmethane dyes such as CV are easily photodegraded by semiconductor catalysts. We designed the CV to photodegrade by transfer of photoinduced electrons from the QD; this process activates the QD PL. The activation photon only needs higher energy than the QD bandgap, so photoactivation does not require UV illumination, which is typically necessary to induce molecular structural changes. As a result, the activation light can be visible or infrared. A QD-CV PBS solution (QD concentration 0.2 μM, CV-to-QD ratio = 3.5, denoted as QD-3.5CV) was prepared. Conjugation of CV to QDs reduced their PL QY to 45% of the initial value (Figure 1c). The QD-CV was irradiated by green light (532 nm, 0.3 mW/ cm2), and the PL was monitored for 120 min (Figure 1d). During this irradiation, PL of the QD-CV recovered to 84% of its initial value at 120 min. No spectral shift was observed during either PL quenching by the CV conjugation or PL activation by the green irradiation. As a control, the unconjugated QD solution was irradiated by green light, which caused no change in absorption or PL (Figure S3c,d). The QD-3.5CV solution kept in darkness showed no change in absorption or PL (Figure S3e,f). The absorption spectrum of QD-3.5CV showed combined features of the QD and CV. Upon exposure to green light, the absorption feature from CV (peak at 590 nm) decreased over time (Figure 1e); after 120 min it had decreased to 12%. Further irradiation with green light induced photodarkening of the QD PL. From 120 to 480 min of the irradiation, the PL intensity dropped to 38% of the initial value, with no spectral shift (Figure 1f). During irradiation, the QD-CV showed photoactivation followed by photodarkening. The QD-CV emission states can be categorized into three stages: (1) during which the QD PL is quenched by the CV conjugation; (2) during which the QD is photoactivated and the PL is recovered; and (3) during which the QD PL is photodarkened and again nonemissive (Figure 1g). During stage 1, PL quenching of QD is attributed mostly to the electron transfers from QD to CVs. The lowest unoccupied molecular orbital (LUMO) energy level of CV is −4.1 eV vs vacuum level, and can therefore readily accept electrons from the conduction band (CB) of the QD.25 During stage 2, the conjugated CV is photodegraded as the QD catalyzes the conversion. Photocatalytic degradation of CV has been observed in many systems.26 When triphenylmethane dyes such as CV undergo photodegradation, they are typically converted into
■
RESULTS AND DISCUSSION PL Modulation of Quantum Dot Crystal Violet Conjugates (QD-CVs). QDs can have large absorption coefficients and the absorption wavelengths can be flexibly tuned. Such QD absorption characteristics was exploited to initiate the QD PL modulation, in which the QD was the photoinduced electron source and the conjugated CV was the PL modulator. CdSe/CdS/ZnS (Core/Shell/Shell) QDs (6.6 ± 0.8 nm) (Figure S1a) were prepared by overcoating CdSe cores (4.2 ± 0.4 nm) with 2 monolayers of CdS, then with 1.4 monolayers of ZnS by following previously reported methods with slight modifications (Supporting Information).20,21 The QDs were surface-modified using a sulfonate-terminated surface ligand (left side in Figure 1g) following a protocol previously reported (Experimental Section).22 QDs dispersed in phosphate-buffered saline (PBS) showed the first excitonic absorption peak at 590 nm, PL peak at 615 nm, and PL quantum yield (QY) > 30% (Figure 1a). The hydrodynamic size was 7.5 ± 0.6 nm (Figure S1c) and the QD surface had negative surface charge (zetapotential −51 ± 5 mV; Figure S1d). CV showed an absorption maximum at 590 nm (Figure 1a). The QD solution was mixed with CV in PBS, and quickly formed QD-CVs by electrostatic attractions between the negatively charged QD surface ligands 7605
DOI: 10.1021/jacs.7b02530 J. Am. Chem. Soc. 2017, 139, 7603−7615
Article
Journal of the American Chemical Society
Figure 2. Photodegradation of CVs induced by QD. (a) 1H NMR and FAB-mass spectra confirming the CV and BDBP. (b) Time-dependent absorption spectra showing the conversion of CV into BDBP with the isosbestic point at 462 nm. Inset: enlarged spectra at the BDBP absorption range around 380 nm. (c) Absorbances of BDBP at 376 nm and CV at 590 nm at different irradiation times in b spectrum. (d) Terephthalate radical assays for QD-CV (red) and unconjugated QD (black) under light irradiation (532 nm, 3 mW/cm2). (e) Blocked generation of hydroxyl radicals by the addition of electron scavenger MV. (f) Inhibited photodarkening of QD-CV by the addition of 0.5% v/v ME.
QD valence band. BDBP is also transparent in the visible range. As the result, BDBP cannot quench the QD PL, so the conversion of CV to BDBP activates the QD. The methyl viologen (MV) could act as a competing inhibitor for the CV receiving electrons.29 The BDBP-producing experiment was repeated using excess MV; in this case the CV conversion to BDBP was significantly impeded (Figure S7b), confirming that the electron transfer facilitates the conversion from CV to BDBP. During stage 3, QD-CV showed PL photodarkening. During the green light irradiation, QD-CV generated significantly more radicals than the unconjugated QD; this result was confirmed by the higher fluorescence signal in the terephthalate assay (Figure 2d). Disodium terephthalate was added to the QD-CV solutions (1 mM). Upon the irradiations of visible light (532 nm, 3 mW/ cm2) for 0, 30, 70, 120, or 170 min, aliquots were withdrawn and treated using sodium hydroxide. PL intensities at 438 nm were measured from the aliquots using excitation at 323 nm. We speculate that radical species create QD surface traps that quench the QD PL. As the electrons were transferred to CVs, the remaining holes may have oxidized water molecules or hydroxide ions to hydroxyl radicals. A control experiment using excess MV was performed to block the electron transfers to CV; the hydroxyl radical generation was significantly impeded (Supporting Information) (Figure 2e). Another control experiment was performed in which excess ME was added as a radical scavenger during stage 3, and the photodarkening was monitored (Supporting Information); the ME completely blocked the photodarkening, and the QD PL remained unchanged upon
benzophenone derivatives. The photodegradation pathways involve oxygen molecules.26 We repeated the QD-CV photoactivation experiment in nitrogen-bubbled solution, but observed no PL activation (Figure S4). Another control experiment was performed using excess of β-mercaptoethanol (ME) as a radical scavenger; no inhibition of CV photodegradation was observed (Figure S5). The results suggest that reactive oxygen species do not significantly contribute to the CV photodegradation of QDCVs. To confirm the benzophenone derivative 4,4′-bis(dimethylamino)benzophenone (BDBP), which is a photodegraded product from CV, a mixture of CV and CdSe powder in PBS was irradiated by green light (532 nm, 3 mW/cm2) for 10 h and the products were characterized by NMR spectroscopy and mass spectrometry (Supporting Information). The photodegraded product from CV, BDBP, was detected (Figure 2a, Figure S6). During the photodegradation experiment, absorption spectra were measured to follow the conversion of CV to BDBP (Figure 2b). The CV absorption gradually decreased concomitantly with the emergence of BDBP absorption that has a peak at 376 nm27 with the isosbestic point at 463 nm. Decrease in the CV absorption peak at 590 nm and concurrent increase in the BDBP peak at 376 nm were plotted (Figure 2c). The control without the CdSe showed no photoreaction (Figure S7a). BDBP cannot accept electrons from the QDs because benzophenone derivatives have higher LUMO levels than do triphenylmethane dyes.28 BDBP should also be a poor hole acceptor because its highest occupied molecular orbital (HOMO) is lower than the 7606
DOI: 10.1021/jacs.7b02530 J. Am. Chem. Soc. 2017, 139, 7603−7615
Article
Journal of the American Chemical Society
Figure 3. Characterization of PL modulation of QD-CVs. (a) Time trend of PL intensity for QD-35CV (red) and the control QD (black) under light irradiation (532 nm, 3 mW/cm2). (b) PL decay over time in unconjugated control QD (black), QD-CV before photoactivation (red), after photoactivation (green), and after photodarkening (blue). (c) Time trend of normalized PL intensity for QD-17.5CV under green light irradiation at fluences of 3 mW/cm2 (black), 6 mW/cm2 (red), and 12 mW/cm2 (blue). (d) Time trend of normalized PL intensity for QD-3.5CV (black), QD10.5CV (red), and QD-17.5CV (blue) under light irradiation (532 nm, 0.6 mW/cm2). Inset: first 50 min.
the fast component showed drastic changes upon the photoactivation and subsequent photodarkening. To the contrary, the slow component was much less sensitive to such modulations. The electron transfer channel opened by CV conjugations was fast enough to prevail over the band-edge emission, which explains the efficient PL modulations upon the CV conjugation and subsequent photoactivation and photodarkening stages. The slow component is attributed to emissions mediated by the shallow traps which provided a temporary reservoir for the excited state shuttling between the traps and the band edge and eventually decaying across the band edge with slower rate than the direct band edge recombination.31 Transition behavior of QD-CVs from stage 1 via 2 to 3 is dependent on fluence rate F and CV-to-QD conjugation ratio. QD-CVs (CV-to-QD ratio = 17.5) in PBS were irradiated by green light at F = 3, 6, and 12 mW/cm2 (Supporting Information) (Figure 3c). As F increased, the QD PL activated increasingly rapid in stage 2. High photon flux should accelerate the electron transfers from QD to CV and also the degradation of the CVs. In addition, as F increased, the transition to stage 3 became increasingly rapid: the photodarkening started at 56, 42, and 30 min respectively at 3, 6, and 12 mW/cm2. The photodarkening resulted from the photogenerated radical species, which should be also dependent on F. PL duration is defined for QD-CVs as the time for which the PL intensity is larger than half of the maximum intensity in stage 2. The PL durations were 93, 76, and 52 min for F = 3, 6, and 12 W/cm2. High photon flux induced rapid photoactivation and photodarkening, which resulted in short PL duration. The effect of CV-to-QD ratio on the stage transition behavior was investigated. QD-CVs of conjugation ratios of 3.5, 10.5, and 17.5 were prepared and the PL transitions from stage 1 via 2 to 3 were monitored under green light irradiation (532 nm, 0.6 mW/ cm2) (Figure 3d). Increase in CV-to-QD ratio resulted in the
continued light irradiation (Figure 2f); this result directly indicates that radical species are involved in the photodarkening during stage 3. PL intensity of QD-35CV (CV-to-QD ratio = 35) changed during irradiation with green light (532 nm, 3 mW/cm2) for 150 min (Figure 3a). For the unconjugated control QD, the PL intensity was almost unchanged over time except for the slight photobrightening which may originate from photoinduced rearrangement of surface molecules or thermal annealing effect of surface atoms.30 Upon CV conjugation, the PL of the QD-CV decreased to 7% of the unconjugated initial. Upon light irradiation, the PL intensity recovered to 91% of the initial at 65 min; i.e., the PL dynamic range was a factor of 13. After 65 min, the QD-CV showed photodarkening and the PL decreased to 15% of the initial at 150 min. This change demonstrates that QD-CV can show complete one cycle of PL turn-on and turn-off. Following stages 1, 2, and 3 of the QD-CV, PL decays were investigated along with the unconjugated QD control (Figure 3b; Supporting Information). Two-exponential fittings were made and the parameters were found for the four samples (Table S1). The unconjugated QD showed average PL lifetime ⟨τ⟩ = 14.7 ns, which decreased to 3.7 ns upon conjugation to QD-CV during stage 1. The radiative rate kr decreased to 73% of the unconjugated initial, whereas nonradiative rate knr increased by 5.4 times (Table S1); the difference in these rates suggests that CV conjugation opens new nonradiative channels and also alters the emission pathway. Upon photoactivation during stage 2, ⟨τ⟩ increased to 9.5 ns, then upon photodarkening dropped to 3.5 ns during stage 3. During stage 2, both kr and knr values returned to close to those of the initial QD. During stage 3, kr decreased to 66% of the unconjugated initial and knr increased by a factor of 5.7. Despite the disparate quenching mechanisms, stages 1 and 3 showed similar changes in kr and knr. For the PL decay parameters fitted by two-exponential components (Table S1), 7607
DOI: 10.1021/jacs.7b02530 J. Am. Chem. Soc. 2017, 139, 7603−7615
Article
Journal of the American Chemical Society
Figure 4. Multicolor photopatterning of the QD-CV film. (a,b) PL images of photoactivated RQD-CV film (a) and GQD-CV film (b). A TEM grid was used as a shadow mask. Scale bar, 100 μm. (c) A scheme of multicolor photopatterning of the film composed of RQD-CVs and GQD-CVs. Irradiation of red light (30 mW/cm2) with the TEM grid on selectively activates the RQD red patterns. Subsequent irradiation by blue light with the TEM grid off activates the GQD green patterns. For the RQD-CVs, the subsequent blue irradiation photoactivates the RQD-CVs in the background area, but photodarkens the RQD-CVs in the square patterns that were preactivated by red irradiation. (d) PL image of the RQD-CV and GQD-CV mixture film before irradiation. (e) PL image of the RQD-CV and GQD-CV mixture film after the red irradiation. (f) PL image of the mixture QD-CV film after the subsequent blue irradiation. Scale bar, 100 μm. (g) PL spectra of RQD (red line) and GQD (green line) samples in PBS buffers. (h) PL spectra measured in regions i and ii of the QD-CV mixture film in image f.
the remaining CVs conjugated to QDs may facilitate photodarkening by promoting the generation of surface traps. To investigate the photomodulation of QD-CV at singleparticle level, fluorescence correlation spectroscopy (FCS) studies were performed. Briefly, as QDs are conjugated by CVs, QD-CVs mainly showed inhomogeneous quenching of the PL; however, the inhomogeneous quenching was accompanied by slight homogeneous quenching that dimmed the QD PL
slowed activation in stage 2. For QD-3.5CV showed over 8 times faster PL increase than QD-17.5CV. Increase in the number of CVs conjugated to a QD should slow its activation by increasing the required number of CV photodegradation events. Transition time to stage 3 decreased from 93 min to 67, and 63 min for QD3.5CV, QD-10.5CV, and QD-17.5CV. QD-CVs with larger conjugation ratios, QD-10.5CV and QD-17.5CV, entered the photodarkening stage 3 faster than QD-3.5CV. During stage 3, 7608
DOI: 10.1021/jacs.7b02530 J. Am. Chem. Soc. 2017, 139, 7603−7615
Article
Journal of the American Chemical Society
Figure 5. Photoactivation and photodarkening of QD-CVs in cells. (a) HeLa cells coincubated with 50 nM QD-CV. (i) Phase-contrast image taken after 2 h of coincubation. After coincubation, cells were irradiated by blue light (11 mW/cm2) and the photoactivation and subsequent photodarkening were monitored by capturing fluorescence images (ii) 0 min, (iii) 20 min, (iv) 1 h 10 min, (v) 4 h 30 min, and (vi) 10 h after irradiation. Scale bars, 20 μm. (b) Control coincubation experiment with unconjugated QD. (i) Phase-contrast image taken after 2 h of the coincubation. After the coincubation, cells were irradiated by blue light (11 mW/cm2). No photomodulation was observed. Fluorescence images (ii) 0 min, (iii) 20 min, (iv) 1 h 10 min, (v) 4 h 30 min, and (vi) 10 h after irradiation. (c) Fluorescence intensities from the QD-CV coincubated cells and from the unconjugated QD coincubated cells over time. White dotted circles: regions of interest.
of photoactivated square patterns than in the masked dark regions. As a control, GQD-CV film was irradiated with red light for 8 h (610 nm, 30 mW/cm2); this film showed no QD PL activation because GQD did not absorb the excitation (Figure S8). Photopatterning was also applied to multicolor QD-CV film that had been obtained by mixing the RQD-CV and GQD-CV solutions together and drop-casting the resulting mixture. Two experiments were performed (Figure 4c): (i) red light irradiation with a TEM grid on for 8 h and (ii) subsequent blue light irradiation for 5 h with the TEM grid off. The multicolor QD-CV film showed no PL before the photoactivation (Figure 4d). After the red light irradiation, the red patterned image was shown (Figure 4e) because GQD-CVs were not activated by the red irradiation. After the additional blue irradiation with the TEM grid off, the red fluorescence pattern changed to green (Figure 4f). For those patterned regions, the blue light irradiation induced photodarkening of RQD-CVs, and concurrently induced photoactivation of GQD-CVs. The background showed yellow fluorescence because RQD-CVs and GQD-CVs were photoactivated concurrently. Figure 4g shows the PL spectra of the RQD and GQD in solutions. The square pattern area and background area in Figure 4f were respectively spectrally resolved (Figure 4h). QD-CVs successfully demonstrated simple and selective multicolor photopatterning. Because QDs have large absorption coefficients, the photopatterning can be relatively rapid under mild light irradiation. Intracellular PL Modulation of QD-CVs. The successful photopatterning using QD-CVs prompted us to observe photoactivation and photodarkening in an intracellular environment. The photomodulation of QD-CVs in a cellular in vivo environment can be particularly useful because visible light or
especially in the case of large number of CV conjugations per QD. Upon irradiation, most of QD-CVs under inhomogeneous quenching were revived by photoactivation. Subsequent photodarkening showed less-efficient inhomogeneous quenching. Details will be discussed later along with Figure 6. Multicolor Photopatterning of QD-CV Films. QD-CVs offer a unique QD conjugate system that exploits dyes that degrade as a result of photoabsorption by the QDs. If the activation relies on the photoabsorption by modulating molecules near the QD, different modulating molecule should be required for selective activations in multiplexed or multicolor systems. Multicolor QD-CVs can use the same modulating molecule CV and can be selectively activated by choosing the excitation wavelength. To demonstrate these advantages of QDCVs and to illustrate their applicability in solid state systems, multicolor QD-CV photopatterning was demonstrated. In comparison to microcontact patterning of QDs, such photopatterning can be advantageous as offering high-resolution and real-time on-demand patterns.32 Two kinds of CdSe/CdS/ZnS QDs were prepared: redemitting QD (RQD, λem = 615 nm; i.e., the same as the sample used for the experiments above) and green-emitting QD (GQD, λem = 560 nm; Figure S1b). RQD-35CV and GQD-35CV aqueous solutions were prepared. Two QD-CV films were prepared (details in Experimental Section). The RQD-CV or GQD-CV solution was drop-cast onto the glass substrate and dried. TEM grids were placed as shadow masks, and the two QDCV films were irradiated by blue light for 5 h (470 nm, 11 mW/ cm2). Both RQD-CV and GQD-CV films showed square photopatterned images under a fluorescence microscope (Figure 4a,b). The blue light induced photoactivations of RQD-CV and GQD-CV films. Fluorescence was seven times higher in regions 7609
DOI: 10.1021/jacs.7b02530 J. Am. Chem. Soc. 2017, 139, 7603−7615
Article
Journal of the American Chemical Society
Figure 6. Fluorescence correlation spectroscopic analysis of QD-CVs. (a) Autocorrelation curves of 0.3-μM unconjugated QD and QD-CV solutions in PBS buffer (red: unconjugated QD, blue: QD-1.8CV, green: QD-3.5CV, dark yellow: QD-7CV, orange: QD-17.5CV, brown: QD-35CV). Dashed black lines: curves fitted to eq 1. (b−e) Plots of count rate (CR) (b), number of QD (N) (c), brightness per particle (BPP) (d), and off-fraction of QD blinking (B) (e) versus the CV to QD ratios of the QD-CV samples. Insets in b, c, and d show the Stern−Volmer plot respectively for CR, N, or BPP versus CV to QD ratio. (f, g) Time trends in the QD-7CV autocorrelation curve under green irradiation (532 nm, 30 mW/cm2) during the stage 2 photoactivation period of 0−16 min (f) and during the stage 3 photodarkening period of 16−51 min (g). Red: fitted curves. (h−k) Plots of CR (h), N (i), BPP (j), and B (k) of QD-7CV vs irradiation time during stages 2 and 3. Dotted lines: of CR (h), N (i), BPP (j), and B (k) of unconjugated QDs. Red and gray shading: the photoactivation (stage 2) and photodarkening (stage 3) periods, respectively.
intracellular environment showed photoactivation and photodarkening behaviors that were very similar to the photopatterning in films. As a control, the experiment was repeated using unconjugated QDs (Figure 5b). The internalized QDs remained constantly bright over time. A representative region was selected for the QD-CV case and for the control (dotted circles in Figure 5), and the fluorescence image intensity was followed over time. The plots followed the same trend as for the QD-CV buffer solution under light irradiation (Figure 3a). Fluorescence Correlation Spectroscopic Analysis of PL Modulation of QD-CVs. FCS allows high temporal resolution for fast photophysical processes in the range of nanoseconds to milliseconds. QD-CVs were excited at 3.5 μW using a 405 nm laser through a water immersion 60× objective lens (Supporting
light at longer wavelengths can be used for the modulation and can minimize adverse effects arising from phototoxicity. The 50nM RQD-CV solution used for the photopatterning in Figure 5 was coincubated with human cervical epithelial (HeLa) cells for 2 h (Experimental Section). Phase-contrast and fluorescent microscope images were taken as the cells were illuminated under blue light (11 mW/cm2) (Figure 5). Initially, the cells were only visible under phase contrast, and no detectable fluorescent image was obtained. As the QD-CVs were photoactivated by illumination, QD-CVs that had internalized into the cells glowed by QD fluorescence with increasing brightness over time until it reached the maximum intensity at 270 min. The photoactivation increased the PL intensity contrast by a factor of 10. Further continued illumination led to photodarkening of the QD-CVs, so the cellular image at 600 min was quite dark. QD-CVs in the 7610
DOI: 10.1021/jacs.7b02530 J. Am. Chem. Soc. 2017, 139, 7603−7615
Article
Journal of the American Chemical Society
Figure 7. Photoactivation and photodarkening of single QD-CVs. (a) (top) Ensemble-averaged (n = 500−1200) PL intensity changes over time during photoactivation and photodarkening periods: GQD-3.5CV under 0.5 W/cm2 irradiation (first column), RQD-3.5CV under 0.5 W/cm2 (second column), GQD-21CV under 0.5 W/cm2 (third column), RQD-21CV under 0.5 W/cm2 (forth column), GQD-21CV under 3.3 W/cm2 (fifth column), and RQD-21CV under 3.3 W/cm2 (last column). Irradiation wavelength was 405 nm. (middle) Five representative single QD-CV PL intensity change over time. (bottom) Time changes in average (n = 24) PL intensity. (b) Single-QD PL intensity trajectories of three representative unconjugated GQDs and three representative unconjugated RQDs under irradiation at 405 nm (0.5 W/cm2).
blinking, τB is the PL off-state relaxation time, and κ is the structure parameter. The amplitude of FCS curves can represent the number of QDs, N. Count rate CR [photons/s; Hz] represents the brightness of QD-CV. The relationship between N and CR during the PL modulation was examined. Using 0.3μM QD-CV (CV-to-QD ratios of 0, 1.8, 3.5, 7, 17.5, and 35) in PBS buffer solutions, FCS studies were performed to follow the PL modulation of QD-CVs at nearly single-particle levels and to characterize the change in blinking dynamics. The unconjugated QD solution showed the decay in the ∼2 ms lag time range, which results from diffusion behavior. During stage 1 of the QDCVs, as the CV to QD ratio was increased, QD-CVs showed an additional decay-pattern in the ∼30 μs lag time range (Figure 6a); this additional decay is attributed to a change in the QD blinking dynamics. The CR decreased from 223 (unconjugated QD) to 156 (QD-1.8CV), 91 (QD-3.5CV), 44 (QD-7CV), 20 (QD-17.5CV), and 8 kHz (QD-35CV) (Figure 6b), which is described well by the Stern−Volmer plot (Figure 6b, inset) and
Information). The resulting excitation volume was ∼1.7 fL. The obtained FCS curves were fitted using a modified autocorrelation function: ⎡ ⎛ τ ⎞⎤ B G(τ ) = GD(τ )⎢1 + exp⎜ − ⎟⎥ ⎢⎣ (1 − B) ⎝ τB ⎠⎥⎦ −1/2 −1 ⎛ 1⎛ τ⎞ τ ⎞ = ⎜1 + ⎟ × ⎜ 1 + 2 ⎟ N⎝ τD ⎠ κ τD ⎠ ⎝ ⎡ ⎛ τ ⎞⎤ B ⎢1 + exp⎜ − ⎟⎥ ⎢⎣ (1 − B) ⎝ τB ⎠⎥⎦
(1)
where GD(τ) is the autocorrelation function by a threedimensional Gaussian diffusion model, τD is the diffusion time of QD in the observation volume, N is the number of QDs in the observation volume, B is the PL off-state fraction in the QD 7611
DOI: 10.1021/jacs.7b02530 J. Am. Chem. Soc. 2017, 139, 7603−7615
Article
Journal of the American Chemical Society
scopic set up equipped with a laser excitation (405 nm continuous-wave laser, 0.5 W/cm2) for GQD-3.5CVs and RQD-3.5CVs in PBS (Figure 7). About 500 individual GQDCVs were detected in the whole CCD area (512 × 512 pixels), and their PLs were averaged (Figure 7a top row). The time evolution of the averaged PL resembled that of ensemble QDCV PL in solution. The higher F in the single QD set up greatly accelerated the speed of photomodulation. The photoactivation was completed within a few minutes and the photodarkening was mostly completed in 20 min. Similarly, PLs from ∼1200 individual RQD-CVs in the entire CCD area were averaged and recorded over time. These showed slightly slower photomodulation; photoactivation was complete at 10 min and the photodarkening took over an hour. In comparison to RQDs, the GQDs have an electron wave function that is more stretched out to the surface because of the higher confinement effect, and thus seem to have more effective electron transfer to the conjugated CVs. Electrons in CB of GQD also have the larger energy level offset to the LUMO of CV.34 Under the same excitation condition, GQD-21CV and RQD-21CV samples were investigated, where the photomodulation was greatly accelerated (Figure 7a, two middle panels of top row). The photoactivation time of GQD-3.5CV was 3 min 40 s and that of GQD-21CV was as short as 16 s. F was increased to 3.3 W/cm2 and the GQD21CV and RQD-21CV samples showed even further accelerated photomodulations with the photoactivation completed within 1 min (Figure 7a, two right panels of top row). At F > 3.3 W/cm2, the PL duration of QD-CV shortened significantly, which made the QD-CV emission bursting at a stochastic time point. Changes in PL over time were recorded for individual QDCVs (Figure 7a, middle row). The photoactivation time varied greatly; some QD-CVs were immediately activated, but others were not activated until >10 min. The time for activation was quite stochastic. During activated period, the trajectories deviated from the binary on or off-state, and may show multiple PL levels which suggests that, the simple two-state model may not be sufficient to describe the photophysics of QD-CVs. This may result from the heterogeneity within single QD-CV conjugated by the multiple CVs and the disparate local environment of the CVs. The individual PL durations were also stochastic, so the PL from individual QD-CVs was highly asynchronous. As the result, multiple QD-CVs within the diffraction-limited area are not likely to show synchronized photoactivation and photodarkening. This minimized temporal overlap between adjacent PL probes is well-suited for superresolution imaging. Change of average PL was plotted for 24 QD-CVs (Figure 7a, bottom row), of which individual trajectories had been followed (some shown in the middle row); this trend was quite similar to the averaged PLs from hundreds of QD-CVs from the entire CCD area (top row). This similarity verifies that the individual trajectories are representative. Single-QD emission time-trends were also obtained for unconjugated GQDs and RQDs; three representative trajectories were shown each type (Figure 7b). They showed continuous emissions with more binary blinking on and off-blinking behaviors than that of QD-CVs. Multicolor Super-Resolution Localization of QD-CVs. QDs constitute a bright and easily multiplexed nanofluorophore class, but they have not been fully exploited for super-resolution imaging. The native blinking of QDs can be used for this process.35 However, native QDs typically show a rather short offstate, so the task of resolving multiple QDs is difficult. PL blueshift of CdSe/ZnS core/shell QDs under high-power irradiation
also accords with the results of the steady-state PL measurements (Figure S3). N also decreased from 58 (unconjugated QD) to 56 (QD-1.8CV), 49 (QD-3.5CV), 34 (QD-7CV), 25 (QD17.5CV), and 20 (QD-35CV) (Figure 6c). However, the decrease in N did not fully account for the CR decrease, as revealed by the nonlinearity of the plot (Figure 6c, inset); this result suggests that some QD-CVs were partially quenched upon CV conjugation. Brightness per particle (BPP) was obtained as CR/N. BPP decreased from 3.87 (unconjugated QD) to 2.80 (QD-1.8CV), 1.86 (QD-3.5CV), 1.28 (QD-7CV), 0.80 (QD17.5CV), and 0.37 kHz (QD-35CV) (Figure 6d); this trend roughly followed the Stern−Volmer plot (Figure 6d, inset). As CVs being conjugated to QDs, a majority of the QD-CVs showed inhomogeneous quenching of the PL; QD-7CVs showed the nearly half “off” and QD-35CVs had the two-thirds “off”. However, the inhomogeneous quenching was also accompanied by a homogeneous quenching that dimmed the QD PL (or BPP), especially as the CV ratio increased. The off-fraction B increased from 0.14 (unconjugated QD) to 0.18 (QD-1.8CV), 0.32 (QD3.5CV), 0.39 (QD-7CV), 0.43 (QD-17.5CV), and 0.53 (QD35CV) (Figure 6e). The conjugated CVs seem to have provided additional shallow trap sites which opened new channels for the QD blinking “off” state. Unconjugated QDs had B = 0.14, which is similar to other reports (Table S2).33 During stage 1, the unconjugated QD buffer solution had CR = 225 kHz, N = 58, BPP = 3.9 kHz, and B = 0.14, which changed to CR = 43 kHz, N = 33, BPP = 1.3 kHz, and B = 0.39 after conjugation to QD-7CV (Figure 6b−e). To analyze the photoactivation, the QD-7CV buffer solution was irradiated with green light (532 nm, 30 mW/cm2) and the FCS curves were monitored over time during the stage 2 (irradiation 0−16 min) (Figure 6f). G(0) dropped continuously during the irradiation; this trend indicates that N increased during photoactivation. During the transition from stage 2 to 3, the G(0) decrease slowed, stopped, and began increasing. During stage 3 (irradiation 16−51 min), G(0) increased over time; this trend accords with the photodarkening (Figure 6g). The temporal changes of CR, N, BPP, and B during the photoactivation and photodarkening (from stage 2 to 3) were plotted (Figure 6h,i,j,k). CR increased from 43 to 183 kHz during stage 2, then decreased back to 90 kHz during stage 3. On a relative scale, CR changed from 100 (before the CV conjugation) to 19 (stage 1), 81 (stage 2), and 40 (stage 3); these changes are similar to the change in steady-state ensemble PL (Figure 3a). From the initial unconjugated state to stage 1, 2, and 3, N changed from 58 to 33, 50, and 39. Upon CV conjugation, about half of the QD population underwent complete quenching, and most of them were revived by photoactivation. Subsequent photodarkening attained somewhat less-efficient inhomogeneous quenching. BPP changed from 3.9 to 1.3, 3.7, and 2.3 kHz. This trend indicates that many individual QDs underwent partial PL modulation during stages 2 and 3. B changed from 0.14 to 0.39, 0.17, and 0.4. Photoactivation seemed to have eliminated most of the shallow traps generated by the CV conjugation. The CV-conjugated state (stage 1) had B = 0.39, which is similar to that in the photodarkened state (stage 3), B = 0.4 (Table S3). Single Particle PL Analysis of QD-CVs. The distinction between stages 2 and 3 is made from the observed change in the ensemble QD-CV PL; photoactivation and photodarkening may occur at the same time. To observe the individual behaviors of QD-CVs, single-QD emission studies were performed using a prism-type total internal reflection fluorescence (TIRF) micro7612
DOI: 10.1021/jacs.7b02530 J. Am. Chem. Soc. 2017, 139, 7603−7615
Article
Journal of the American Chemical Society
Figure 8. Two-color colocalization of QD-CVs within a diffraction-limited area. (a) PL intensity trace from two GQD-21CVs (G1, G2) in a diffractionlimited area under 405 nm excitation (3.3 W/cm2). Insets: wide-field CCD images; each was taken when the corresponding QD-CV was photoactivated. Scale bars in insets, 200 nm. (b) Centroids of G1 and G2 obtained by fitting a Gaussian function. (c) Enlargement of the dotted box in b; bars indicate standard deviation in the x and y direction, of the centroid distributions. Calculated distance between G1 and G2 was 46.6 nm. (d−f) Similar to a−c, PL intensity trace, two CCD images, centroids, and the standard deviations are shown for two RQD-21CVs (R1, R2). Distance between R1 and R2 was 165 nm. (g) Simultaneous localizations of the four QD-CVs in a synchronized coordinate system.
QD-CVs to precision finer than the diffraction limit. To confirm the potential of using QD-CVs as multicolor super-resolution probes, a mixture of GQD-CVs and RQD-CVs was used for photomodulated localization microscopy using the single-QD TIRF set up. The color signals from GQD and RQD were separated using a setup with filters and beam splitters (Scheme S1). Coordinations of the green and red channels were synchronized by second-order polynomial warp transformation procedures using GQD and RQD polymer coencapsulates prepared by a method previously reported (Supporting Information).38 1 × 1 CCD pixels that showed both green and red QD emissions were selected, and the signals of each color channel (green: Figure 8a,b,c; red: Figure 8d,e,f) were separated and followed over time. Individual PL bursts from different GQD-CVs could be detected (Figure 8a). In a single CCD pixel, the centroids of two GQD-CVs were determined by twodimensional Gaussian fits, which yielded two resolved locations (G1, G2) separated by 46.6 nm, with standard deviations (x direction, y direction) of (11.5 nm, 8.2 nm) for G1 and (12.0 nm, 9.1 nm) for G2 (Figure 8b,c). In the same CCD pixel, the red channel signal was also followed (Figure 8d); coincidentally, PL modulations from two RQD-CVs (R1, R2) were observed; they were separated by 165 nm, with standard deviations (32.2 nm, 20.8 nm) for R1 and (48.9 nm, 46.6 nm) for R2 (Figure 8e,f). The positions of the G1, G2, R1 and R2 were plotted in a synchronized coordinate system (Figure 8g). The GQD-RQD distances were 164 nm for G1-R1, 302 nm for G1-R2, 147 nm for G2-R1 and 266 nm for G2-R2. The Abbe diffraction limits for our system are 233 nm for GQD and 256 nm for RQD. Distances of G1-G2 (46.6 nm), R1-R2 (165 nm), G1-R1 (164 nm) and G2R1 (147 nm) are all below these limits, but the four QDs of two colors within the diffraction-limited area were successfully localized and distinctly resolved at nanoscale distances. Multicolor super-resolution localizations using QD-CVs have been demonstrated under mild excitation (405 nm, 3.3 W/cm2). When necessary, multicolor QD-CVs can be sequentially activated on-demand as demonstrated in photopatterning
(19 kW/cm2) has been used to localize PL of a small number of QDs to image microtubules at-12 nm resolution by exploiting different rates of QD-bluing among individual QDs.36 This process requires physical deterioration of inorganic QDs, so excessively high power excitation is unavoidable. Doped QDs have been reported for super-resolution imaging.18 The doped QD strategy is not generally applicable to many QDs, and multiplexing is limited. QD-CVs can be used as super-resolution probes that offer simple multicolor imaging under mild photoexcitation. Conventional photoswitchable fluorescent dyes require more than one laser excitation for multicolor super-resolution imaging. For example, conventional multicolor fluorescent organic probes composed of activator-and-reporter dye pair need different kinds of activators that require different wavelengths for activation.37 Addition of chemical species such as thiols is also often necessary to enhance the photoswitching. Multicolor QD-CVs can avoid these complications by being modulated by a single laser without the need for extra chemicals. The photomodulation of QD-CV is limited to one cycle from off to on and back to off. However, such a single cycle can be advantageous for some super-resolution imaging, for example to avoid blurring by multiple photoactivations of a single probe at different positions by drift or movement. In addition, the high brightness of QD should provide sufficient signal to enable accurate localization of the target structures within the single photoactivation event. The PL duration of the single photoactivation and photodarkening cycle in QD-CVs can be tuned by controlling the CV conjugation ratio and excitation condition. The single QD-CV PL duration can be significantly longer (>1 min) than conventional dyes (
