Subscriber access provided by University of Massachusetts Amherst Libraries
Article
High-Density Super-Resolution Localization Imaging with Blinking Carbon Dots Hua He, Xu Liu, Shan Li, Xiaojuan Wang, Qian Wang, Jiqiang Li, Junying Wang, Hao Ren, Baosheng Ge, Shengjie Wang, Xiao-Dong Zhang, and Fang Huang Anal. Chem., Just Accepted Manuscript • DOI: 10.1021/acs.analchem.7b03567 • Publication Date (Web): 04 Oct 2017 Downloaded from http://pubs.acs.org on October 4, 2017
Just Accepted “Just Accepted” manuscripts have been peer-reviewed and accepted for publication. They are posted online prior to technical editing, formatting for publication and author proofing. The American Chemical Society provides “Just Accepted” as a free service to the research community to expedite the dissemination of scientific material as soon as possible after acceptance. “Just Accepted” manuscripts appear in full in PDF format accompanied by an HTML abstract. “Just Accepted” manuscripts have been fully peer reviewed, but should not be considered the official version of record. They are accessible to all readers and citable by the Digital Object Identifier (DOI®). “Just Accepted” is an optional service offered to authors. Therefore, the “Just Accepted” Web site may not include all articles that will be published in the journal. After a manuscript is technically edited and formatted, it will be removed from the “Just Accepted” Web site and published as an ASAP article. Note that technical editing may introduce minor changes to the manuscript text and/or graphics which could affect content, and all legal disclaimers and ethical guidelines that apply to the journal pertain. ACS cannot be held responsible for errors or consequences arising from the use of information contained in these “Just Accepted” manuscripts.
Analytical Chemistry is published by the American Chemical Society. 1155 Sixteenth Street N.W., Washington, DC 20036 Published by American Chemical Society. Copyright © American Chemical Society. However, no copyright claim is made to original U.S. Government works, or works produced by employees of any Commonwealth realm Crown government in the course of their duties.
Page 1 of 8
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
Analytical Chemistry
High-Density Super-Resolution Localization Imaging with Blinking Carbon Dots Hua He1, Xu Liu1, Shan Li1, Xiaojuan Wang1, Qian Wang1, Jiqiang Li1, Junying Wang2, Hao Ren1, Baosheng Ge1, Shengjie Wang1, Xiaodong Zhang2,*, and Fang Huang1,* 1 State Key Laboratory of Heavy Oil Processing and Center for Bioengineering and Biotechnology, China University of Petroleum (East China), Qingdao 266580, China 2 Department of Physics and Tianjin Key Laboratory of Low Dimensional Materials Physics and Preparing Technology, School of Science, Tianjin University, Tianjin 300350, China.
ABSTRACT: Molecular fluorescence blinking provides a simple and attractive way to achieve super-resolution localization via conventional fluorescence microscopy. However, success in super-resolution imaging relies heavily on their blinking characteristics. We here report easily prepared and photostable nanoparticles—carbon dots (CDs) —with desirable fluorescence blinking for highdensity super-resolution imaging. The CDs exhibit a low duty cycle (~ 0.003) and high photon output (~ 8000) per switching event, as well as show much higher resistance to photobleaching than Alexa 647 or Cy5 typically used in single molecule localization microscopy. The stable blinking of CDs allows to perform high-density localization imaging at a resolution of 25 nm by sequentially recording the particle positions. The CD-based super-resolution imaging is further demonstrated by rendering CD-stained tubular peptide self-assemblies, CD-packed clusters with well-defined patterns, and CD-stained microtubules in a cell. Furthermore, this method has been validated as a valuable tool to detect the clustering and distribution of protein receptors in the plasma membrane that are not discerned with normal fluorescence imaging.
The ability to image subcellular structures beyond the optical diffraction limit has revolutionized our view of cells since the recent advent of super-resolution microscopies that are achieved either by patterned illumination such as stimulated emission depletion (STED),1 or by image reconstruction such as photo-activated localization microscopy (PALM)2 and stochastic optical reconstruction microscopy (STORM).3-5 Particularly, PALM/STORM that rely on single molecule localization almost reach the ultimate spatial resolution for molecular mapping in a cell, determined by the currently available fluorescent probes.6 Conceptually, PALM/STORM sequentially localize fluorophores that are randomly switched between fluorescent (“on”) and dark (“off”) by modulating the activation and excitation lasers, and then assemble all positions over thousands of cycles for a super-resolution image. Those photo-switchable fluorophores include fluorescent proteins, organic dyes or pairs of dyes.7-9 STORM can also be achieved in its simplest form only using a conventional fluorescent dye (e.g. Alexa Fluor 647 and Cy5) and a single continuously illuminating laser for both activation and excitation.6,10,11 Such an implementation makes super-resolution imaging methods accessible to most labs in a wide research community. In addition to fluorescent proteins and dyes, inorganic quantum dots (QDs) hold a great promise for single molecule/particle imaging in biological system due to their remarkable photostability and brightness.12 In particular, QDs are also useful candidates for super-resolution fluorescence imaging. This is mainly benefited from their natural blinking phenomenon observed in most QDs.13 Actually, the QD blinking-based superresolution was first reported in 2005.14 Nevertheless, the typical
QDs represented by CdSe/ZnS share a blinking behavior with a short-time “off” and long-time “on” states. Such a characteristic limits the number of QDs that can be distinguished within a subdiffraction region,15-17 and thus imposes a great constraint on super-resolution imaging of practical biological specimen with densely packed targets. Recently, the blinking of QDs has been also exploited in another imaging technique termed super-resolution optical fluctuation imaging (SOFI) to enhance spatial resolution,18-20 but the SOFI image is created by calculating the statistical moments of each pixels rather than localizing single emitters, thus lacking the ability to provide single-molecule information for quantifying target molecules. Herein, we introduce a new class of fluorescent nanoparticles made of carbon materials (also known as carbon dots, CDs) with burst-like blinking behavior well suited for super-resolution localization imaging. The CDs are emerging nanomaterials in recent years that show great promise in bio-imaging because of high biocompatibility and extremely small size (< 5 nm).21 Although so far there has been few reports on the use of CDs for super-resolution imaging with STED and SOFI,22,23 the actual emission of single CDs has never been extracted and resolved. The CDs presented here exhibit a blinking behavior with low duty cycle and high photon output per switching event as well as show robust photostability similar to QDs. This allows acquiring an image stack that contains sparse subsets of single spots in each frame for sequential localization. By sequentially recording their positions, highly densely packed particles can be isolated with a localization precision of 10 nm (corresponding to a spatial resolution of ~ 25 nm), and then assembled into a high-density super-resolution image. The super-resolution imaging is
ACS Paragon Plus Environment
Analytical Chemistry
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
demonstrated by rendering CD-stained tubular peptide self-assemblies, CD-packed clusters with well-defined patterns, and CD-stained microtubules in a cell. Applying this method, the heterogeneous distribution and aggregation of proteins hidden under conventional fluorescence microscopy can be clearly visualized at the molecular level, allowing the analysis of nanoscale clustering and arrangement of protein receptors in the plasma membrane. EXPERIMENTAL SECTION Synthesis and characterization. CDs were prepared as described in our previous works.24-26 Briefly, carbon black (0.2 g) was refluxed with 6 mol L-1 nitric acid (50 mL) for 24 h. The resultant suspension was cooled to room temperature (20-25 oC) and then centrifuged. After the pellet was discarded, the supernatant was heated and dried. The obtained reddish-brown solid was resuspended in ultra-pure water, and then successively ultra-filtered with 3 kDa and 30 kDa cut-off membranes. The CDs with molecular weights (MWs) between 3-30 kDa was finally retained. The CD samples were diluted for optical characterization. Ultraviolet-visible (UV-Vis) spectra were collected using a Shimadzu UV-2450 spectrophotometer. Fluorescence spectra and 3D fluorescence spectra were recorded on the Horiba FluoroMax-4 spectrometer. Transmission electron microscopic (TEM) and High-resolution TEM (HRTEM) images were recorded on a JEM-2100 electron microscope at 200 kV. Atomic force microscopic (AFM) images were obtained using a Nanoscope IV Multimode AFM (Digital Instruments, Santa Barbara). Preparation of peptide self-assemblies and CD-packed nanopatterns. The purified KIIIIK (KI4K) peptides were dissolved in ultra-purified water (18.2 MΩ) to create a solution at a concentration of 16 mM. After the peptides were completely dissolved by sonication for about 5 min, the solution was adjusted to pH 3.0 using dilute HCl and then incubated for 14 days at room temperature before further characterization. The CDstained peptide assemblies were prepared by adding CDs into the solution containing peptide assemblies for 15 min incubation before fluorescence imaging. For CD-packed nanopatterns, the silicon wafer was first etched by electron beam lithography (JBX-6300FS, JEOL, Japan) into the nanopatterned stamp that has a square lattice of 400 nm × 400 nm with a separation of 50 nm in trench width. The stamp was dip-coated with the CD solution and then placed face to face on a coverslip before fluorescence imaging. Conjugation of CDs to goat anti-mouse IgG. CDs in methanol (30 µL, 50 mg/mL) was allowed to react with EDC (100 mg/mL) and NHS (100 mg/mL) at room temperature for 30 min. The activated CDs was mixed with goat anti-mouse IgG secondary antibody (2.5 mg/mL in PBS) and then reacted for more than 4 h. The amount of methanol was kept below 4-5% of total reaction volume. The solution was repeatedly ultra-filtered using a filtering membrane with a molecular weight (MW) cutoff of 100 kDa against PBS buffer. The purified IgG-CD conjugates were finally stored at 4 oC for further use. Fluorescence correlation spectroscopy (FCS) and dynamic light scattering (DLS, ZEN3600, Malvern Instruments, UK) were used to evaluate the conjugation between CDs and IgG. Cell Culture and immunostaining. Human cervical cancer HeLa cells were cultured with high Dulbecco’s Modified Eagle’s medium (DMEM) supplemented with 10% fetal bovine serum (FBS) in a humidified incubator at 37 °C where CO2 level was kept constant at 5%. To immunostain microtubules, the
Page 2 of 8
cells were washed with PBS once and fixed in a 1:1 acetone/ methanol solution for 10 min. After three washes with PBS, the cells were blocked by incubation with 6% BSA and 10% normal goat serum in PBS for 2 h. After removing the blocking buffer, the cells were incubated with mouse anti-alpha tubulin primary antibody diluted to 1:200 (2.5 µg/mL) in 6% BSA at 4 oC overnight. The cells were then washed three times with the washing buffer (0.2% BSA and 0.05% Triton X-100 in PBS) for 10 min per wash. After that, CD-labeled goat anti-mouse IgG secondary antibody (~ 2 µg/mL) was added to the cells in 6% BSA and incubated for 2 h. The cells were washed again three times with the washing buffer and one time with PBS for 10 min per wash and stored in PBS before imaging. To immunostain chemokine receptor CCR3 in the plasma membrane, the pcDNA4/ToCCR3 or pcDNA4/To-EGFP CCR3 plasmids were first transfected into HeLa cells using Lipofectamine 2000. After transfection, cells were further cultured for 24 h. The cells were then fixed in 4% paraformaldehyde (PFA) in PBS for 15 min. The fixed cells were first blocked with 15 % fetal bovine serum (FBS) in 10 uL PBS for 30 min, and then 50 uL mouse antihuman CCR3 primary antibody diluted to 1:200 (2.5 µg/mL) was added for incubation overnight at 4 oC. After washed three times by PBS and then blocked with 15% FBS in 500 uL PBS for 30 min. After that, 50 uL of CD-labeled goat anti-mouse IgG secondary antibody (~ 2 µg/mL) was added and incubated for 1h. Subsequently, the cells were washed three times with PBS. The CD-stained cells on coverslips were stored in PBS at 4 oC before imaging. Fluorescence imaging and single molecule analysis. All single molecule/particle imaging on coverslips or cell membranes were performed on a Nikon total internal refection fluorescence microscopy (TIRFM). The imaging of dye molecules was conducted in PBS solution (7.4) containing 100 mM β-mercaptoethanol to enhance their photoswitching performance. For microtubule imaging inside a cell, the incidence angle was adjusted slightly below the critical angle such that the excitation light illuminated 1-2 µm deep into the sample. CDs, CdSe/ZnS QDs and Cy3 were excited using 532 nm laser in conjunction with 575 nm dichroic mirror and 590 nm long-pass filter; Cy5 and AF647 were excited using 635 nm laser in conjunction with 640 nm dichroic mirror and 692/40 nm band-pass filter. The excitation density was about 20 W cm-2. The image was taken by an Andor iXon 897 EMCCD (16 µm/pixel) using 100 × objective combined with 1.5 × magnification changer lens, and the resultant effective dimension for each pixel was about 106 nm. The electron multiplying (EM) gain of EMCCD was set to 300 and the pixel readout rate was set to 1 MHz at 16 bit. The fluorescence intensity (analog digital units, ADU) was converted to photons by using the electrons/ADU conversion factor (DCF), EM gain and quantum efficiency (QE) as follows: photons=ADU × DCF /(QE ×gain). Here, DCF was 21.62 and QE for 550 nm-700 nm light was about 92%, as provided by the manufacturer. For single molecule analysis, the integrated signal was collected from 7 × 7 pixels2 (742 × 742 nm2) around the maximum of a single spot. To avoid the crosstalk of multiple particles into an intensity trajectory, any fluorescent spot within five pixels of another fluorescent spot was omitted from analysis. Fluorescence blinking events in the time traces were identified by identifying signal changes with a magnitude that is greater than 6 times the standard deviation of the background fluctuations of the fluorescence time trace.
ACS Paragon Plus Environment
Page 3 of 8
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
Analytical Chemistry
Reconstruction of super-resolution images. We first acquired a time series of fluorescence images. The super-resolution images were then reconstructed with Thunder-STORM,27 an ImageJ plugin. This plugin provides a complete tool for image pre-processing, particle localization, post-processing analysis and visualization of data. The pre-processing step can be also performed in MATLAB before fitting and localizing single particles with Thunder-STORM. Here, the images were preprocessed by performing background subtraction and Gaussian filtering to remove the noise. The intensity maxima, which were greater than or equal to all values within an 8-connected neighborhood, were identified, and single particle localization was performed by a least-squares fit of an assumed two-dimensional Gaussian point spread function (PSF), yielding new coordinates for the location of the particle, with a position uncertainty The super-resolution image was rendered by combining all localized particles in the image stack using a density estimation approach based on average shifted histograms provided in Thunder-STORM. Cluster analysis. The protein clustering was analyzed by Ripley’s K function on the basis of their localization data.28-30 In the reconstructed image, a tested region of 2 × 2 μm2 was typically selected. Ripley’s K-function was calculated as ( ) = ∑ ∑ , , where A is the image area, N is the number of total localizations in the area, r is the spatial radius for the K-function calculation and δij is the distance between the i-th and the j-th localizations. if δij is less than r, the value will be 1, otherwise δij = 0. This counts the number of other points encircled by concentric rings centered on each point. The linear transformation of K(r) was used to interpret the spatial randomness ( ) − =
( )
− . The amplitude of
L(r) − r would be zero for particles with random distribution, and positive for clustering particles. Edge-effects were negated by weighting edge points and cropping image edges after the calculation. The values of L(r) generated by each particle were used to produce a cluster map by interpolating a surface plot with L(r) as the z-axis. Then a binary cluster map was generated through a defined L(r) threshold. The percentage of points which satisfied L(r) − r > 0 was 60%, and the L(r) threshold was set at 40% of the maximum value of L(r) from the plot. Finally the information of clustering could be extracted from the binary map, such as the cover area and number of clusters. All calculations and image processing were performed in MATLAB. RESULTS AND DISCUSSION Structure and optical characterization. The synthesis of CDs was conducted on an oxidation of carbon black in HNO3, and carboxyl-coated carbon nanomaterials were obtained as described in our previous works.24-26 Differing with previous products comprising single-sheet graphene (< 3000 Da), the final products we used here were obtained between 3000-30,000 Da through the successive dialysis. Transmission electron microscopy (TEM) image shows that they are carbon nanoparticles of about 4.5 nm in diameter with a narrow size distribution (Figure 1a and b). High-resolution TEM (HRTEM) image reveals the existence of lattice. The lattice spacing is determined to be about 0.21 nm, which is comparable to the (100) facet of graphite.21 Atomic force microcopy (AFM) manifests that those particles have heights between 3 and 8 nm, further implying a multilayer structure (Figure 1c and d). We thus name them as CDs instead of graphene QDs. They show wide optical absorption ranging from 200 to 700 nm with the first exciton peak at about
510 nm (Figure 1e). Upon excitation, CDs exhibit a sharp fluorescence emission peak at about 590 nm. The excitation-emission matrix indicates two main excitation regions with centers of 350 nm and 510 nm (Figure 1f).
Figure 1. Morphology and optical characterization of CDs. a) TEM and HRTEM (inset) images; b) Particle size distribution determined from HRTEM images; c) AFM image; d) 3D AFM image shows a height range of 3-8 nm; e) UV-vis absorption and fluorescence spectra of CD solution; f) 3D fluorescence map of CDs.
Single particle imaging and blinking. To evaluate the fluorescence of CDs at the single particle level, organic dyes including Cy3, Cy5 and AF647, and commercial CdSe/ZnS QDs emitted at 605 nm were also analyzed as comparisons (Figure 2 and S1-6). Here, CDs, Cy3 and QDs were excited with 532 nm laser, while AF647 and Cy5 were excited with 635 nm laser. AF647 and Cy5 were chosen for comparison because they were reported as being superior dyes in STORM imaging. To highlight the single molecule/particle imaging quality, Figure 2a-c give fluorescence images of CDs, AF647 and QDs without any image filtering processing. CDs show bright fluorescent spots and high signal/noise ratios comparable to AF647, but they are inferior to CdSe/ZnS QDs in term of brightness. For example, 1 s exposure time was required to obtain clear fluorescence images for CDs and AF647, but only 100 ms for CdSe/ZnS. Besides, a distinct difference was observed between their emission behaviors. Figure 2d-f and S1-5 show typical time traces of single spots for different fluorophores. The singe-step transition to the dark state implies that single spot in each frame was produced from single particle emission. As observed, single CDs and dye molecules exhibit the blinking/bleaching process with long times in the dark state, while CdSe/ZnS QDs display an opposite blinking behavior with long-lived “on” and short-lived dark states. Moreover, dye molecules could hardly maintain fluorescence emission after 600 s of illumination due to their photobleaching. But, the CDs and QDs were observed to have blinking behaviors during the whole illumination period of 30 min. This difference can also be intuitively observed with the videos of CDs, Cy3 and QDs (Video S1-3). To quantitatively compare their performance for super-resolution imaging, we
ACS Paragon Plus Environment
Analytical Chemistry
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
calculated the duty cycle (fraction of time in the “on” state) during the first 600 s of illumination and mean photons per switching event for each type of fluorophore from more than 100 trajectories (Figure 2g, h). Their photon-number distributions of each switching event are given in Figure S1-5. CDs have a photon output (~ 8000 per switching event) close to AF647 or Cy5, and a low duty cycle down to ~ 0.003. The high photon number and low duty cycle are essential to achieving high resolution in STORM imaging. For CdSe/ZnS QDs, although they have much higher photon number than CDs and dye molecules, the too high duty cycle (~ 0.7) means that most of particles fluoresce simultaneously at any given moment as shown in Figure S6a. This makes densely packed particles difficult to be resolved by successively recording their positions. The other key factors to limit high-density localization are photostability and switching cycle. In the case of Cy3, despite its relatively low duty cycle and high photon output, most fluorescent spots disappeared within 300 s (Figure S6b). The rapid photo-bleaching lacks the ability to provide efficient switching cycles for highdensity localization. Figure 2i further shows the number changes of single molecules/particles for each type of fluorophore during 30 min illumination. It can be seen that the CDs show comparable photostability to CdSe/ZnS QDs, and much higher than dye molecules. The high photostability of CDs was also demonstrated by continuous illumination of high-power xenon lamp in the ensemble solution (Figure S7). Although single CDs have switching cycles (about 2-3) close to AF647 or Cy5 during the first 600 s of illumination, the high photostability can substantially increase switching cycles by extending a total acquisition time. Because of the stable and burst-like fluorescence, the imaging of CDs brought up each frame with an extremely sparse field of fluorescent spots over a large population of dark particles in the image sequence (Figure S6c). In fact, the stochastic bursts of fluorescent spots in each frame could be observed during a continuous illumination of up to 1 h until no spots appeared (data not shown), although 30 min was chosen in this work as the analysis time window for different fluorophores. Such blinking feature is quite favorable for superresolution localization of high-densely packed particles.
Figure 2. Single molecule/particle fluorescence imaging and quantitative analysis. Fluorescence images and single molecule/particle intensity trajectories for a, d) CDs, b, e) AF647 and c, f) CdSe/ZnS QDs. Exposure time, 1 s for CDs and AF647, and 100 ms for CdSe/ZnS QDs. Scale bar, 5 µm. g-i) On-off duty cycle values, mean photons per switching event, and the number fraction during continuous illumination for each type of fluorophore, respectively.
Since the first observation of blinking phenomenon of semiconductor QDs in 1996,13 the blinking mechanism has not been
Page 4 of 8
well elucidated until now. The most widely accepted explanation is that blinking arises from the escape of the photo-excited electrons (or holes) to the trap states with exponentially distributed trap depths on the surface of nanocrystals or the surrounding substrates. In the resultant charged or charge-separated nanocrystals the fluorescence is quenched via non-radiative Auger recombination until the electrons (or holes) tunnel back to the nanocrystals, rendering it bright again.31 For carbon-based fluorescent nanomaterials that have been recently developed, such as CDs and graphene QDs, their emission was supposed to be associated with the much more complex surface groups and environments than those in semiconductor QDs.32,33 In our previous works,24-26 we revealed the existence of –COOH, –OH and epoxy groups on the CDs. Following the theory explaining the QD blinking, although the fluorescence mechanism of CDs remains unclear to date, we postulate that the rich bond structures contribute to a large amount of surface states that provide broad and deep energy wells to accept the ejected electrons (or holes). Once captured in those “hotbeds”, the returning probability would be low, resulting in long-lived “off” states. The long-duration dark states make sense in a low QY of approximately 5% for CDs we measured with quinine sulfate as a reference (QY=56%), far lower than those of semiconductor QDs (up to 80%).34,35
Figure 3. Principle of super-resolution imaging with CDs. a) Schematic concept of super-resolution reconstruction by fitting single spots with a 2D Gaussian function as illustrated with a point array. b) The intensity histogram of a single CD. c) Distribution of localization precisions from more than 3000 CDs. d, e) The centroids of a single QD during 30 min before and after the correction of the system drift.
Particle localization and super-resolution imaging. Due to the optical diffraction, the fluorescence signal of a single spot spreads several pixels, that is, point spread functions (PSF). When each frame in the image stack is projected at maximum across time t as illustrated in Figure 3a, the fluorescence signals overlap to produce a diffraction-limited image similar to those obtained with conventional fluorophores. For CDs, however, all particles in a PSF area are scarcely possible to appear simultaneously because of the low burst probability. We can thus perform fitting of single spots in each image with a two-dimensional (2D) Gaussian function, localize their centroid positons and finally superimpose all positions into a super-resolution image. The localization uncertainty (or precision) is described36 /
as ∆ = + , where σ is the standard deviation (SD) of the PSF, a is the pixel size (106 nm) in the image, b is the background noise, and N is the number of photons. Single CDs can emit efficient number of photons at mean intensity of ~3,000 photons/s, which give a Gaussian intensity distribution (Figure 3b). Compared to the molecular signal, the background noise is negligible. By statistical analysis, the averaged
ACS Paragon Plus Environment
Page 5 of 8
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
Analytical Chemistry
localization uncertainty of particles is 5.5 nm (Figure 3c). Nevertheless, due to thermal and mechanical instability, the drift of the sample stage exists unavoidably during the continuous recording, further exacerbating the localization uncertainty. In our microscopy, the setup is equipped with a Perfect Focus System (PFS, Nikon, Japan), which automatically corrects z-axis drift in real time. No substantial defocusing was observed in all the videos we acquired. In x and y directions, however, a significant drift occurred, as evidenced by the broad and skewed position distribution in Figure 3d, by sequentially recording the positions of the identical spot during 30 min. Single CdSe/ZnS QDs were here used as the fiducial marker for the tracking of position movement because they maintain “on” states over most of the frames in the stack. The position uncertainty from the system drift is described15 as ∆ = + , where σx and σy are the SDs of the drifts in the x and y direction. After the correction, the SDs of the system drift could be improved from 27 to 8.5 nm (Figure 3e). All corrected positions of CDs using fiducial markers should be used for the superimposition into a super-resolved image, giving the overall localization uncertainty ∆ =
∆χ
+∆
. Such localization uncer-
tainty was here calculated to 10 nm, which corresponds to a spatial resolution of about 25 nm.16
Figure 4. Super-resolution localization of CDs coated on a coverslip a) z-projected image at maximum. b) Super-resolution reconstructed image. The small regions are enlarged on the respective bottoms. c, d) The intensity profiles of line regions as indicated with dash lines in the enlarged image.
To confirm the effective super-resolution localization, we first imaged CDs directly coated on a coverslip. In each frame, the fluorescent spots were identified and then localized. Those too weak or deformed spots giving localization precisions of more than 10 nm were rejected. Figure 4a, b present the timeprojected image at maximum across 600 frames and the reconstructed super-resolution image, respectively. Small regions marked with the square boxes are enlarged on the respective bottoms for detail observation. The comparative analysis of the two images demonstrates a drastic improvement in spatial resolution. Those spots that are unresolvable in the diffraction-limited image are clearly isolated. Figure 4c presents a representative line profile of two nearby spots showing only a wide peak within a diffraction-limited region. After reconstruction, two
well-separated peaks are evidently observed (Figure 4d), establishing an accurate distance of 200 nm. To validate the capability to render a fine structure, we imaged KI4K peptide self-assemblies stained with CDs. The KI4K is a symmetric amphiphilic peptide which can self-assemble in aqueous solution into long nanotubes with a homogenous diameter distribution.37,38 The atomic force microscopy (AFM) images in Figure 5a and S8 show that their diameters are about 250 nm. Because KI4K self-assemblies contain rich positive amino groups on the surface, they are easily labeled with negative CDs via the electrostatic attraction. Owing to the high-density accumulation of CDs on the assemblies, the fluorescence image time-projected from only 300 frames gives a tubular feature (Figure S9a), but with a wide diameter of about 750 nm. After super-resolution rendering, a clear visual improvement of tubular self-assemblies is presented (Figure S9b). The images of 3 selected peptide self-assemblies are enlarged in Figure 5b and c for clearer observation. In the super-resolution image, they show a narrower nanotube shape with a sharp edge, defining a mean width of about 250 nm in consistent with that measured by AFM (Figure 5d). The resolution is improved by approximately 3 fold in comparison with that in the diffraction-limited image. Furthermore, such super-resolution localization imaging permits viewing the nanometer distribution of CD-labeled targets at high density, revealing the ultrastructure features. This capability is demonstrated by visualizing CD-packed clusters with well-defined patterns on the silicon wafer, which is etched by electron beam lithography into a square lattice of side 400 nm with a separation of 50 nm in trench width (Figure 5e, f). The nanometer clusters are unrecognizable under conventional fluorescence microscopy but can be clearly rendered by localizing blinking CDs.
Figure 5. High-density super-resolution imaging of CD-stained samples. a) AFM image of KI4K peptide self-assembly. Inset shows a schematic diagram of the peptide self-assembled nanotube. b, c) The time-projected image and super-resolution reconstructed image across 300 frames of CD-stained peptide-assemblies. d) From top to bottom, the line profiles of the regions as indicated in a, b and c, respectively. e) From left to right, SEM image of the nanopatterned silicon wafer, the time-projected image and superresolution reconstructed image across 300 frames of CD-packed patterns, respectively. f) The enlarged SEM image of four blocks and super-resolution reconstructed image of four CD-packed blocks.
Super-resolution imaging of subcellular structures. Applying blinking CDs to cell imaging, we first imaged microtubules immunostained with primary antibodies and CD-labeled
ACS Paragon Plus Environment
Analytical Chemistry
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
secondary antibodies. The conjugation of CDs to secondary antibodies was confirmed by fluorescence correlation spectroscopy and dynamic light scattering (Figure S10), and the conjugates did not exhibit significant particle aggregation before staining as evidenced with uniform intensity distribution and burst-like blinking (Figure S11 and Video S4). Figure 6a and b show the conventional fluorescence image and super-resolution image of CD-stained microtubules in the same region of cell, respectively. The super-resolution image shows a drastic improvement in the resolution of the microtubule network. In the region where microtubules cannot be resolved in the conventional image, individual microtubules are clearly separated by super-resolution imaging with blinking CDs (Figure 6c-e). The average width of microtubules from the Gaussian fit (FWHM) is determined to 60 ± 6 nm, which is consistent with previous measurements of the antibody-labeled microtubule width from TEM or STORM imaging.4,39 In addition to imaging microtubules inside a cell, the application of super-resolution fluorescence imaging to cell membrane is exciting because the heterogeneous distributions and organizations of most membrane proteins are not yet adequately understood.40,41 For example, G protein-coupled receptors (GPCRs) are the largest family of membrane proteins that control many cellular signaling pathways by undergoing spatial organization and arrangement in different structures such as clusters.42 The clusters are intrinsically protein aggregates with nanoscale sizes that cannot be discerned by conventional fluorescence microscopy, but believed to work as a signaling platform involved in complex cellular process. As a proof of concept, we transfected chemokine receptor CCR3 as an example of GPCRs into HeLa cells,43 The validity of the transfection protocol was examined with EGFP-fused CCR3, showing the heavy expression and accumulation of CCR3 on cell membrane (Figure S12). The CCR3 was immunostained with primary antibodies and CD-labeled secondary antibodies, and then imaged under TIRFM. The nonspecific interaction of CDs with cells was eliminated by adding 15% FBS in each step, so that no CDs were observed on cell membrane in the absence of primary antibodies except for weak cell auto-fluorescence under epi-illumination (Figure 6f). The specific immunostaining of cells with CDs presents sporadic spot distribution in each frame (Video S5), but the severe burst-like blinking implies the existence of a larger population of dark CDs. The mean fluorescence intensity determined from 10 cells indicates higher signal than those cells without CDs under the same TIR illumination, further confirming their specific binding of CDs with cell membrane (Figure S13). The time-projected image at maximum intensity across 600 frames gives a bright and rich distribution of CDs and then depicts a typical picture of cell membrane (Figure S14). Nevertheless, it appears to be difficult for one to analyze the distribution pattern of proteins due to the optical diffraction limit. By the rendering method we developed with CDs, the reconstructed image shows a substantial improvement in resolution (Figure 6g). A magnified view of the super-resolution image plotted using each localization as a scatter plot is shown in Figure S15 for clearer observation. Spot-distributed patterns with more or less clustering are clearly visible but smear out in the diffraction-limited image. To confirm the existence of clustering, Ripley’s K-function analysis was employed to estimate spatial randomness on the base of localization data (See cluster analysis in Experimental Section). In a tested region of 2 × 2 µm2 (Figure 6h), the r value corresponding to the maximum of L(r) – r represents the average cluster diameter (Figure 6i). In the color-encoded map, red represents highly clustered regions
Page 6 of 8
whereas blue denote low values for clustering (Figure 6j). By setting a threshold, the resulting binary map can reveal the area and number of clusters (Figure 6k). By analyzing 30 regions from 10 cells, the average clustering diameter of CCR3 is about 190 ± 20 nm with clustering range scaled up to 520 ± 30 nm. The cover area of CCR3 clusters is about 4.5 %, and 51% of total CCR3 is participating in clustering. To verify the results, we conducted the PALM imaging and clustering analysis of EFGP-CCR3 (Figure S16). The cluster size is about 180 ± 20 nm with clustering range scaled up to 490 ± 20 nm. These clusters have a cover area of 3.6%, and about 49% of EGFP-CCR3 participating in clustering. By comparison, only a very slight increase is observed in the case of CD-stained cells likely due to antibody-decorated receptors and a very small fraction of CD agglomeration. However, this will not affect the demonstration that our method can be used a powerful tool to characterize the clustering and distribution of proteins in the nanoscale range.
Figure 6. Super-resolution imaging of CD-labeled structures inside a cell and on the plasma membrane. Scale bar, 10 µm. a) The conventional fluorescence and b) the super-resolution image of microtubules labeled with CDs; c, d) The enlarged view of small regions in a) and b). e) The intensity profile of line region as indicated in d). f) The cell membrane stained with CDs in the absence of antiCCR3 primary antibodies. Inset shows the cell auto-fluorescence image of the same field under epi-illumination. g) The super-resolution image of cell membrane immuno-stained with CDs; h) The enlarged 2×2 µm2 region in g). i) Ripley’s K function analysis of CCR3 clustering in the region. j) The interpolated cluster map based on Ripley’s K function analysis. k) The corresponding binary cluster image using a threshold.
CONCLUSION In conclusion, we report a high-density super-resolution localization imaging method with blinking CDs. The CDs are easily prepared and have burst-like fluorescence with low duty cycle, high photon output and high photostability. This feature allows acquiring a dataset for super-resolution localization with a spatial resolution of 25 nm. The high-density imaging capability is demonstrated with improved resolutions by rendering CDstained peptide self-assemblies, CD-packed clusters with welldefined patterns and CD-stained microtubules inside a cell. The present work also provides a valuable tool to detect the clustering and distribution of CD-labelled protein receptors in the plasma membrane at the molecular level.
ACS Paragon Plus Environment
Page 7 of 8
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
Analytical Chemistry
ASSOCIATED CONTENT Supporting Information Imaging movies of CdSe/ZnS QDs, Cy3, CDs, IgG-CDs and CDstained cell, blinking analysis of CDs, AF647, CdSe/ZnS QDs, Cy5 and Cy3, photostability of CDs in the ensemble solution , AFM and super-resolution image of peptide self-assemblies, FCS and DLS characterizations of IgG-CDs, fluorescence image of IgG-CDs, TIRF and PALM image of EGFP-CCR3 transfected cells, the fluorescence intensity analysis of CD-stained cells, the diffractionlimited and super-resolution image of CD-stained cells. The Supporting Information is available free of charge on the ACS Publications website.
AUTHOR INFORMATION Corresponding Author *E-mail:
[email protected] *E-mail:
[email protected]
Notes The authors declare no competing financial interests.
ACKNOWLEDGMENT This work was supported by the Natural Science Foundation of Shandong Province (ZR2014BM028), the National Key Basic Research Program of China (2012CB518000), and the National Natural Science Foundation of China (21273287, 21403300, 21373271, 81471786 and 21773310).
REFERENCES (1) Hell, S. W.; Wichmann, J. Opt. Lett 1994, 19, 780-782. (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, 1642-1645. (3) Rust, M. J.; Bates, M.; Zhuang, X. Nat. Methods 2006, 3, 793-796. (4) Bates, M.; Huang, B.; Dempsey, G. T.; Zhuang, X. Science 2007, 317, 1749-1753. (5) Huang, B.; Wang, W.; Bates, M.; Zhuang, X. Science 2008, 319, 810813. (6) Dempsey, G. T.; Vaughan, J. C.; Chen, K. H.; Bates, M.; Zhuang, X. Nat. Methods 2011, 8, 1027-1036. (7) Bates, M.; Huang, B.; Zhuang, X. Curr. Opin. Chem. Biol. 2008, 12, 505-514. (8) Lippincott-Schwartz, J.; Patterson, G. H. Trends Cell Biol. 2009, 19, 555-565. (9) Patterson, G.; Davidson, M.; Manley, S.; Lippincott-Schwartz, J. Annu. Rev. Phys. Chem. 2010, 61, 345-367. (10) Heilemann, M.; Van De Linde, S.; Schüttpelz, M.; Kasper, R.; Seefeldt, B.; Mukherjee, A.; Tinnefeld, P.; Sauer, M. Angew. Chem., Int. Ed. 2008, 47, 6172-6176. (11) Heilemann, M.; van de Linde, S.; Mukherjee, A.; Sauer, M. Angew. Chem., Int. Ed. 2009, 48, 6903-6908. (12) Michalet, X.; Pinaud, F. F.; Bentolila, L. A.; Tsay, J. M.; Doose, S.; Li, J. J.; Sundaresan, G.; Wu, A. M.; Gambhir, S. S.; Weiss, S. Science 2005, 307, 538-544. (13) Nirmal, M.; Dabbousi, B. O.; Bawendi, M. G.; Macklin, J.; Trautman, J.; Harris, T.; Brus, L. E. Nature 1996, 383, 802-804.
(14) Lidke, K.; Rieger, B.; Jovin, T.; Heintzmann, R. Opt. Express 2005, 13, 7052-7062. (15) Chien, F. C.; Kuo, C. W.; Chen, P. Analyst 2011, 136, 1608-1613. (16) Wang, Y.; Fruhwirth, G.; Cai, E.; Ng, T.; Selvin, P. R. Nano Lett. 2013, 13, 5233-5241. (17) Xu, J.; Tehrani, K. F.; Kner, P. ACS nano 2015, 9, 2917-2925. (18) Watanabe, T. M.; Fukui, S.; Jin, T.; Fujii, F.; Yanagida, T. Biophys. J. 2010, 99, L50-52. (19) Dertinger, T.; Colyer, R.; Iyer, G.; Weiss, S.; Enderlein, J. Proc. Natl. Acad. Sci. U. S. A. 2009, 106, 22287-22292. (20) Gallina, M. E.; Xu, J.; Dertinger, T.; Aizer, A.; Shav-Tal, Y.; Weiss, S. Opt. Nanoscopy 2013, 2, 1-9. (21) Baker, S. N.; Baker, G. A. Angew. Chem. Int. Ed. 2010, 49, 67266744. (22) Leménager, G.; De Luca, E.; Sun, Y.-P.; Pompa, P. P. Nanoscale 2014, 6, 8617-8623. (23) Chizhik, A. M.; Stein, S.; Dekaliuk, M. O.; Battle, C.; Li, W.; Huss, A.; Platen, M.; Schaap, I. A.; Gregor, I.; Demchenko, A. P.; Schmidt, C. F.; Enderlein, J.; Chizhik, A. I. Nano Lett. 2016, 16, 237-242. (24) Wang, X.; Wang, Y.; He, H.; Chen, X.; Sun, X.; Sun, Y.; Zhou, G.; Xu, H.; Huang, F. J. Mater. Chem. B 2016, 4, 779-784. (25) Wang, X.; Sun, X.; He, H.; Yang, H.; Lao, J.; Song, Y.; Xia, Y.; Xu, H.; Zhang, X.; Huang, F. J. Mater. Chem. B 2015, 3, 3583-3590. (26) Wang, X.; Sun, X.; Lao, J.; He, H.; Cheng, T.; Wang, M.; Wang, S.; Huang, F. Colloids Surf., B 2014, 122, 638-644. (27) Ovesný, M.; Křížek, P.; Borkovec, J.; Švindrych, Z.; Hagen, G. M. Bioinformatics 2014, 30, 2389-2390. (28) Owen, D. M.; Rentero, C.; Rossy, J.; Magenau, A.; Williamson, D.; Rodriguez, M.; Gaus, K. J. Biophotonics 2010, 3, 446-454. (29) Gao, J.; Wang, F.; Liu, Y.; Cai, M.; Xu, H.; Jiang, J.; Wang, H. Sci Rep 2015, 5, 9045. (30) Gao, J.; Wang, Y.; Cai, M.; Pan, Y.; Xu, H.; Jiang, J.; Ji, H.; Wang, H. Nanoscale 2015, 7, 2511-2519. (31) Efros, A. L.; Nesbitt, D. J. Nat. Nanotechnol. 2016, 11, 661-671. (32) Wang, L.; Zhu, S. J.; Wang, H. Y.; Qu, S. N.; Zhang, Y. L.; Zhang, J. H.; Chen, Q. D.; Xu, H. L.; Han, W.; Yang, B. ACS nano 2014, 8, 25412547. (33) Zhu, S.; Song, Y.; Zhao, X.; Shao, J.; Zhang, J.; Yang, B. Nano Res. 2015, 8, 355-381. (34) Qu, L.; Peng, X. J. Am. Chem. Soc. 2002, 124, 2049-2055. (35) Dabbousi, B. O.; Rodriguez-Viejo, J.; Mikulec, F. V.; Heine, J. R.; Mattoussi, H.; Ober, R.; Jensen, K. F.; Bawendi, M. G. J. Phys. Chem. B 1997, 101, 9463-9475. (36) Thompson, R. E.; Larson, D. R.; Webb, W. W. Biophys. J. 2002, 82, 2775-2783. (37) Zhao, Y.; Wang, J.; Deng, L.; Zhou, P.; Wang, S.; Wang, Y.; Xu, H.; Lu, J. R. Langmuir 2013, 29, 13457-13464. (38) Zhao, Y.; Deng, L.; Wang, J.; Xu, H.; Lu, J. R. Langmuir 2015, 31, 12975-12983. (39) Weber, K.; Rathke, P. C.; Osborn, M. Proc. Natl. Acad. Sci. U. S. A. 1978, 75, 1820-1824. (40) Stone, M. B.; Shelby, S. A.; Veatch, S. L. Chem. Rev. 2017. (41) Gunzenhauser, J.; Olivier, N.; Pengo, T.; Manley, S. Nano Lett. 2012, 12, 4705-4710. (42) Bethani, I.; Skånland, S. S.; Dikic, I.; Acker-Palmer, A. EMBO J. 2010, 29, 2677-2688. (43) Wang, M.; Ge, B.; Li, R.; Wang, X.; Lao, J.; Huang, F. PLoS One 2013, 8, e65500.
ACS Paragon Plus Environment
Analytical Chemistry
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
Page 8 of 8
For Table of Contents Only
ACS Paragon Plus Environment
8
