We describe the preparation of colloidal quantum dots with minimized hydrodynamic size for single-molecule fluorescence imaging. Compared to conventional quantum dots, these nanoparticles are similar in size to globular proteins and are optimized for single-molecule brightness, stability against photodegradation, and resistance to nonspecific binding to proteins and cells.
Single-molecule imaging is an important tool for understanding the mechanisms of biomolecular function and for visualizing the spatial and temporal heterogeneity of molecular behaviors that underlie cellular biology 1-4. To image an individual molecule of interest, it is typically conjugated to a fluorescent tag (dye, protein, bead, or quantum dot) and observed with epifluorescence or total internal reflection fluorescence (TIRF) microscopy. While dyes and fluorescent proteins have been the mainstay of fluorescence imaging for decades, their fluorescence is unstable under high photon fluxes necessary to observe individual molecules, yielding only a few seconds of observation before complete loss of signal. Latex beads and dye-labeled beads provide improved signal stability but at the expense of drastically larger hydrodynamic size, which can deleteriously alter the diffusion and behavior of the molecule under study.
Quantum dots (QDs) offer a balance between these two problematic regimes. These nanoparticles are composed of semiconductor materials and can be engineered with a hydrodynamically compact size with exceptional resistance to photodegradation 5. Thus in recent years QDs have been instrumental in enabling long-term observation of complex macromolecular behavior on the single molecule level. However these particles have still been found to exhibit impaired diffusion in crowded molecular environments such as the cellular cytoplasm and the neuronal synaptic cleft, where their sizes are still too large 4,6,7.
Recently we have engineered the cores and surface coatings of QDs for minimized hydrodynamic size, while balancing offsets to colloidal stability, photostability, brightness, and nonspecific binding that have hindered the utility of compact QDs in the past 8,9. The goal of this article is to demonstrate the synthesis, modification, and characterization of these optimized nanocrystals, composed of an alloyed HgxCd1-xSe core coated with an insulating CdyZn1-yS shell, further coated with a multidentate polymer ligand modified with short polyethylene glycol (PEG) chains (Figure 1). Compared with conventional CdSe nanocrystals, HgxCd1-xSe alloys offer greater quantum yields of fluorescence, fluorescence at red and near-infrared wavelengths for enhanced signal-to-noise in cells, and excitation at non-cytotoxic visible wavelengths. Multidentate polymer coatings bind to the nanocrystal surface in a closed and flat conformation to minimize hydrodynamic size, and PEG neutralizes the surface charge to minimize nonspecific binding to cells and biomolecules. The end result is a brightly fluorescent nanocrystal with emission between 550-800 nm and a total hydrodynamic size near 12 nm. This is in the same size range as many soluble globular proteins in cells, and substantially smaller than conventional PEGylated QDs (25-35 nm).
The following synthesis procedures involve standard air-free techniques and the use of a vacuum/inert gas manifold; detailed methodology can be found in references 10 and 11. MSDS for all potentially toxic and flammable substances should be consulted before use and all flammable and/or air-labile compounds should be aliquoted into septum-sealed vials in a glove box or glove bag.
1. Synthesis of Mercury Cadmium Selenide (HgxCd1-xSe) Quantum Dot Cores
2. Growth of Cadmium Zinc Sulfide (CdyZn1-yS) Shell
3. Phase Transfer
4. PEG Coating
Figure 2 depicts representative absorption and fluorescence spectra for CdSe nanocrystals, HgxCd1-xSe nanocrystals after cation exchange, and HgxCd1-xSe/CdyZn1-yS nanocrystals after shell growth. The core CdSe nanocrystals have a quantum yield of fluorescence near 15% (including long-wavelength deep-trap emission) but this efficiency drops to less than 1% after mercury exchange, likely due to charge carrier traps introduced through surface atom disruption 9. However the growth of a thin shell of CdyZn1-yS boosts this efficiency to more than 70%, which is largely maintained after transfer to water (50% is typical). In contrast, CdSe/CdyZn1-yS nanocrystals without mercury incorporation lose a substantial fraction of their quantum yield in water unless a thick shell is grown. Thus by incorporating mercury into the core nanocrystal, the small size of the nanocrystal can be maintained (see TEM in Figure 3) without sacrificing brightness. It is important to note that capping with CdyZn1-yS shifts the spectra to the red due to leakage of the electronic charge carriers into the shell material; this shift is around 20-30 nm for CdSe cores 16, and increases with increasing mercury content in the core (up to 100 nm).
The use of a 2-step phase transfer to water is critical for obtaining a homogeneous population of nanocrystals that do not require further size sorting to remove clusters and aggregates. In the first step, the nanocrystals are transferred to DMSO using 1-thioglycerol, which displaces oleylamine on the surface of the nanocrystal. Thioglycerol is then replaced with a linear multidentate polymer, resulting in highly stable particles with a minimal increase in hydrodynamic size resulting from the organic coating (<4 nm contribution to the hydrodynamic diameter). The size-exclusion chromatogram depicted in Figure 4a confirms that the size is similar to that of conalbumin (75 kDa), and after modification with 750 Da amino-PEG, the size is increased to just 12 nm, similar to that of an IgG antibody. PEG modification neutralizes the surface charge, as confirmed in the agarose gel electrophoresis experiment depicted in Figure 4b. We routinely use size-exclusion chromatography and gel electrophoresis for quick characterization of size, size distribution, and surface charge. Dynamic light scattering and zeta potentiometry can also be used, however the scattering cross-section of these ultrasmall particles is very small, and we have found that results from commercial instruments are not reproducible. Figure 5a shows an epifluorescence micrograph of these nanocrystals deposited on a glass coverslip and excited with 545 nm visible light. These nanocrystals are readily observed at the single-molecule level at 30 frames per second with an electron-multiplying CCD camera. Figure 5b shows that the number of fluorescent particles observed in each frame fluctuates over time with continuous excitation; this is due to a combination of blinking and photodegradation. Blinking dominates for the first ~7 min before oxidative photodegradation slowly becomes apparent.
Figure 1. Schematic depiction of the nanoparticle synthesis procedure. (a) Cadmium and selenium precursors react to generate CdSe nanocrystals, which are treated with mercury octanethiolate, inducing partial Cd→Hg cation exchange to yield HgxCd1-xSe ternary alloy nanocrystals. A shell of CdyZn1-yS is then grown on the core using cadmium acetate, zinc acetate, and sulfur. (b) As synthesized, these nanocrystals are coated with nonpolar organic ligands (oleylamine). To solubilize these particles in aqueous buffers, the ligands are replaced with a multidentate polymer ligand, which is covalently coupled to amino-PEG.
Figure 2. Optical properties of HgxCd1-xSe/CdyZn1-yS nanocrystals. (a) Absorption (black) and fluorescence spectra (red) of CdSe nanocrystal cores, HgxCd1-xSe cores after cation exchange, and HgxCd1-xSe/CdyZn1-yS nanocrystals after shell growth. Spectra are offset for clarity (b) Fluorescence spectra of HgxCd1-xSe/CdyZn1-yS with different relative amounts of mercury incorporation. The blue spectrum depicts cores with zero mercury content (x=0, CdSe).
Figure 3. Transmission electron micrograph (a) and particle size distribution (b) of HgxCd1-xSe/CdyZn1-yS nanocrystals, showing an average diameter ± standard deviation of 3.2 ± 0.6 nm.
Figure 4. Hydrodynamic characterization of HgxCd1-xSe/CdyZn1-yS QDs in aqueous solution. (a) Size exclusion chromatogram of nanocrystals coated in a multidentate polymer ligand before (red) and after (blue) conjugation to amino-PEG. Molecular weight protein standards are indicated above the plots. (b) Agarose gel electrophoresis experiment of the QDs in sodium borate buffer (pH ~8.5) before (left) and after (right) conjugation to amino-PEG. The well is marked with an arrow and electrode polarities are indicated on the right, showing that before conjugation the nanocrystals migrate as anionic particles and the PEGylated nanocrystals are electrostatically neutral.
Figure 5. HgxCd1-xSe/CdyZn1-yS QDs adsorbed on a glass coverslip in phosphate buffer, imaged with epifluorescence microscopy. (a) QD image obtained at 33 frames per second. Image is 15 μm x 15 μm. (b) Number of fluorescent QDs per field of view during continuous illumination for 20 min with mercury arc lamp with 545 nm (30 nm bandpass) excitation filter and a 625 nm (20 nm bandpass) emission filter and 100x 1.4 NA objective. Measurements from 3 fields of view were averaged over 20 min at 12.5 frames per second.
Compared to conventional CdSe quantum dots, ternary alloy HgxCd1-xSe nanocrystals can be tuned in size and fluorescence wavelength independently. The size is first selected during the synthesis of CdSe nanocrystal cores, and the fluorescence wavelength is chosen in a secondary mercury cation exchange step, which does not substantially alter the nanocrystal size 9. It is important to allow the purified HgxCd1-xSe nanocrystals to incubate at room temperature for at least 24 hours before capping. This allows some of the weakly adsorbed mercury cations to diffuse into the nanocrystal lattice. Without allowing this process to occur, a second fluorescence band in the near-infrared is often observed due to homogeneous nucleation of HgS nanocrystals from dissociated mercury ions.
In the example shown in this work, we prepared CdSe cores with a size near 2.3 nm, which can be tuned in fluorescence between 550-800 nm after capping by altering the amount of mercury incorporated into the core lattice. With a 2.5 monolayer shell, the final diameter of these QDs was near 3.2 nm, which is essentially the smallest size nanoparticle that we can prepare that is both sufficiently photostable and sufficiently bright for single-molecule imaging (extinction coefficient near 350,000 M-1 cm-1 at 400 nm and quantum yield near 50% in water). These nanocrystals are substantially brighter and more photostable than previously described nanocrystals with comparable sizes that emit over this spectral range (e.g. CdTe, InAs, InP). Like most fluorophores, the fluorescence from these particles at the single molecule level is intermittent (blinking) 5,6.
For some applications, it may be beneficial to use somewhat larger nanocrystals. By using a larger CdSe nanocrystal core, the fluorescence bandwidth is narrower after mercury cation exchange. Typical fluorescence peak widths for HgxCd1-xSe nanocrystals with emission in the 600-650 nm window are 50-70 nm for 2.3 nm cores and 40-50 nm for 3.2 nm cores. Thereby, larger nanocrystals enable a greater capacity for spectral multiplexing. In addition, increasing the size will likewise increase the absorption cross-section of the nanocrystals. Increasing the thickness of the CdS interim shell layer will also increase the brightness, and further prolong the fluorescence stability during excitation. The CdSe core size may be increased simply by extending the duration of the CdSe core synthesis, and monitoring the effective size through UV-Vis absorption spectrophotometry.
We have found that aqueous QDs coated with carboxylic acids are prone to nonspecific adsorption to cells and proteins, and that neutralization of their strong negative charge in physiological buffers is critical for minimizing nonspecific interactions 17. In the examples here, we used short-chain PEG to neutralize the surface charge and maintain stability in water. PEG can be introduced into the polymer backbone either before attachment to the QDs or after coating. Both procedures result in nearly neutral particles, but those first coated with the carboxyl-polymer are substantially smaller, presumably due to improved multidentate interaction with the surface. For complete surface neutralization with PEG, we have found that repeated addition of carboxylic acid activating agents is necessary due to the short half-life of the reactive species. We use DMTMM in the place of more common carbodiimide reagents (e.g. EDC) because of the improved stability of DMTMM in storage and due to improved reaction efficiency in water 18.
Finally, it is important to note that quantum dots and many other types of nanocrystals contain cytotoxic elements 5. Cadmium and mercury ions can affect the normal processes of living cells and organisms and may be carcinogenic 19-21. However the cytotoxicity of conventional CdSe/ZnS nanocrystals has been widely studied and it has been reported that robustly coated nanocrystals with stable organic ligands do not elicit overtly cytotoxic responses compared to their constituent elements, simply because their toxic elements are efficiently sequestered away from oxidizing agents 5. Moreover, for single-molecule imaging applications, toxic effects are unlikely due to the extremely small concentrations used for imaging (typically 1 nM or less) which are orders of magnitude smaller than the onset of detectable toxic effects (50-100 nM). Most of the single-molecule experiments implementing QDs to date have utilized commercially available CdSe/ZnS nanocrystals, which are substantially larger than those described herein. By minimizing the nanocrystal size, the total number of surface atoms per particle and the total number of toxic atoms per particle are substantially reduced, thereby reducing the total potential for toxicological impact. The incorporation of mercury into the nanocrystal is expected to further reduce the toxicity potential, as divalent mercury is known to be less toxic than divalent cadmium in many cell types 19-21.
The authors have nothing to disclose.
The authors would like to thank Dr. Hong Yi at the Emory University Integrated Microscopy Core for electron microscopy imaging. This work was sponsored by NIH grants (PN2EY018244, R01 CA108468, U54CA119338, and 1K99CA154006-01).
Name of the reagent | Company | Catalogue number | Comments (optional) |
Selenium | Sigma-Aldrich | 229865 | |
Tri-n-octylphosphine | Strem | 15-6655 | 97% pure, unstable in air |
Cadmium oxide | Sigma-Aldrich | 202894 | Highly toxic: use caution |
Tetradecylphosphonic acid | PCI Synthesis | 4671-75-4 | |
Octadecene | Alfa Aesar | L11004 | Technical grade |
Hexadecylamine | Sigma-Aldrich | H7408 | |
Diphenylphosphine | Sigma-Aldrich | 252964 | Pyrophoric |
Mercury acetate | Sigma-Aldrich | 456012 | Highly toxic: use caution |
1-Octanethiol | Sigma-Aldrich | 471836 | Strong odor |
Oleic acid | Sigma-Aldrich | W281506 | |
Zinc acetate | Alfa Aesar | 35792 | |
Cadmium acetate hydrate | Sigma-Aldrich | 229490 | Highly toxic: use caution |
Oleylamine | Fisher Scientific | AC12954 | Unstable in air |
Sulfur | Sigma-Aldrich | 344621 | |
Trioctylphosphine oxide | Strem | 15-6661 | 99% |
Pyridine | VWR | EM-PX2012-6 | Anhydrous |
Thioglycerol | Sigma-Aldrich | M1753 | Strong odor |
Triethylamine | Sigma-Aldrich | 471283 | Anhydrous |
Dialysis tubing | Spectrum Labs | 131342 | 20 kDa cutoff |
Centrifugal filter | Millipore | UFC801024 | 10 kDa cutoff |
Monoamino-PEG | Rapp Polymere | 12 750-2 | 750 Da |
DMTMM, 4-(4,6-Dimethoxy-1,3,5-triazin-2-yl)-4-methylmorpholinium chloride hydrate | Alfa Aesar | H26333 | |
AKTAprime Plus Chromatography System | GE HealthCare | ||
Superose 6 10/300 GL chromatography column | GE HealthCare | 17-5172-01 | |
Agarose, OmniPur | VWR | EM-2120 | |
Appendix Synthesis of mercury octanethiolate: Slowly add a methanol solution of mercury acetate (1 eq.) to a stirring solution of 1-octanethiol (3 eq.) and potassium hydroxide (3 eq.) in methanol at room temperature. Isolate the mercury(II) octanethiolate precipitate via filtration, wash two times with methanol and once with ether, and then dry under vacuum. Synthesis of multidentate polymer: Dissolve polyacrylic acid (1 g, 1,773 Da) in 25 ml dimethylformamide (DMF) in a 150 ml three-necked flask and bubble with argon for 30 min. Add an anhydrous solution of cysteamine (374 mg, 4.87 mmol) in 10 ml DMF. At room temperature with vigorous stirring, slowly add anhydrous diisopropylcarbodiimide (DIC, 736 mg, 5.83 mmol) over 30 min, followed by triethylamine (170 μl, 1.22 mmol), and allow the reaction to proceed for 72 hr at 60 °C. Add mercaptoethanol (501 mg, 6.41 mmol) to quench the reaction, and stir for 2 hr at room temperature. Remove DMF via rotary evaporation and isolate the polymer with the addition of a 2:1 mixture of ice-cold acetone:chloroform, followed by centrifugation. Dissolve the polymer in ~5 ml anhydrous DMF, filter, precipitate again with diethyl ether, and repeat. Dry the product under vacuum and store under argon. Determination of CdSe core diameter: From the UV-Vis absorption spectrum determine the wavelength of the first exciton peak (λ, in nm), which is the longest-wavelength peak (e.g. roughly 498 nm for CdSe in Figure 2a), and use the sizing curve of Mulvaney and coworkers 12: Determination of CdSe nanocrystal concentration: From a background-subtracted UV-Vis spectrum of an optically clear solution of CdSe nanocrystals, determine the absorption at 350 nm wavelength. Serial dilutions can be used to determine if the optical absorption is within the linear range of Beer’s Law. The nanocrystal concentration (QD, in M) can be determined by plugging in the nanocrystal diameter (D, in nm), the optical absorption value (A3sa), and the cuvette path length (l, in cm) into the following equation from the empirical correlation of Bawendi and coworkers 13: |