Imaging dynamic cellular events with quantum dots The bright future

Abstract

Semiconductor quantum dots (QDs) are tiny light-emitting particles that have emerged as a new class of fluorescent labels for biology and medicine. Compared with traditional fluorescent probes, QDs have unique optical and electronic properties such as size-tuneable light emission, narrow and symmetric emission spectra, and broad absorption spectra that enable the simultaneous excitation of multiple fluorescence colours.

These properties arise from their semiconducting nature, which exhibits quantum size effects when charge carriers (electrons and holes) are confined in dimensions smaller than a critical size threshold defined by the Bohr diameter¹. Semiconductor nanocrystals that are smaller than the Bohr diameter emit fluorescent light that can be tuned by the particle size across a broad spectrum of UV, visible, near-IR and even mid-IR wavelengths. The semiconducting crystalline nature of these particles also gives rise to large absorption cross-sections, which causes QDs to be 10–100 times brighter than organic dyes and fluorescent proteins (see Figure 1). In addition, these particles are dramatically more resistant to photo-degradation compared with traditional probes. However, QDs are macromolecules that are an order of magnitude larger than organic dyes, which can limit their use in cellular imaging applications that require small labels. In this article, we briefly discuss recent developments in size-minimized QDs, new tagging strategies, as well as methods for QD intracellular delivery and dynamic cellular imaging.

Figure 1.

Optical properties of QDs. (a) Vials containing six different sizes of QDs composed of cadmium selenide (CdSe) dissolved in solution are illuminated with a UV lamp. (b) Fluorescence spectra of QDs depicted in (a). (c) The absorption spectra (blue) and fluorescence spectra (red) of QDs are compared with those of fluorescent proteins (mCherry) and organic dyes (Texas Red), showing the broad absorption spectra of QDs and their narrow and symmetric fluorescence spectra. The relative sizes of these fluorescent labels are depicted next to their optical spectra.

Compact and bright QDs

The large size of current QD probes arises mainly from a thick polymer coating on the particle surface. The nanocrystalline cores of QDs are initially prepared in non-polar organic solvents and coated with non-polar aliphatic ligands. In order to prepare water-soluble colloids, they are coated with an amphiphilic polymer shell (see Figure 2). These formulations are exceptionally stable and biocompatible and have been instrumental in enabling the commercialization of QD bioprobes, but the micelle-like multilayer coating greatly increases the total hydrodynamic size. For example, QDs with a 5 nm crystalline core will balloon in overall size to 15–30 nm after coating. In comparison, most soluble globular proteins are 4–8 nm in diameter.

Figure 2.

Schematic diagrams showing the structures of traditional QDs and size-minimized QDs. Traditional QDs are coated with a thick bilayer of aliphatic ligands and amphiphilic polymers that result in a large overall size. Compact QDs are coated with only a monolayer of bidentate or multidentate ligands that contain either thiols or amine functional groups (red). The multidentate ligands are anchored on the QD surface via either a brush-like monolayer or a flat wrapping conformation.

The first biocompatible QDs were stabilized in water with small thiolate ligands such as mercaptoacetic acid². These particles were exceptionally compact in size, but the co-ordination bond between the ligand and the QD surface is weak, causing rapid degradation of these particles in complex biological conditions. This problem has recently been overcome by re-engineering these organic ligands to increase their binding affinity to the QD surface through multidentate interactions^3–5. As shown in Figure 2, low-molecular-mass hydrophilic polymers can be anchored to the nanocrystal surface through multiple thiol and/or amine moieties to generate highly stable nanoparticles. This success has dramatically reduced the sizes of biocompatible QDs to close to that of globular proteins, opening new opportunities for intracellular imaging and tracking studies.

New tagging strategies

In order to use QDs as fluorescence labels to image and track biomolecules, the chemical linkage between the QD and a biomolecule is critically important. The conventional approach for fluorescent biotagging is to conjugate the fluorophore to an antibody that has a specific affinity for a protein or biomolecule of interest (see Figure 3a). However, the QD has a large surface area available for coupling, which results in multiple antibodies binding to each particle. This results in a bulky structure that can form cross-links between multiple target molecules, prohibiting the imaging of single biomolecules. In addition, most covalent coupling schemes result in random orientations of the antibody on the QD surface, which renders many of the antibodies inactive owing to steric hindrance from the particle surface. The ideal tagging strategy would be to attach a QD to the target biomolecule with precisely controlled orientation through a high-affinity bond with a minimal linker length, at 1:1 or controllable stoichiometric ratios. One promising approach is based on the use of polyhistidine peptides (His-tags), which are frequently appended to the termini of recombinant proteins in order to provide an efficient purification route using metal-affinity chromatography⁶. His-tags bind with high affinity and specificity to bivalent metal atoms such as Ni²⁺ or Zn²⁺ loaded on columns. Because QDs also have metal-rich surfaces, proteins with a His-tag domain efficiently assemble on the QD surface with a well-defined orientation (Figure 3b). New thin-shell QD coating strategies are readily adaptable to this conjugation scheme owing to the accessibility of the nanocrystal surface, and a variety of purification methods can be used to isolate discrete QD–protein conjugates to obtain monovalent tagged proteins⁷. Although more systematic studies are needed, preliminary work from our own laboratory indicates that the His-tag–QD attachment can withstand complex biological environments for prolonged periods of time without significant dissociation.

Figure 3.

Strategies for biomolecular tagging of QDs. (a) Antibody–antigen interactions. The broad diversity of available antibodies enables the production of a broad range of QD bioaffinity probes. Antibody attachment to the QD is often mediated by highly modular lock-and-key streptavidin–biotin interactions that results in bulky conjugates with multiple antibodies per QD. (b) His-tag chelation. His-tags appended to proteins enable site-specific binding to the metal atoms on QD surfaces. (c) Monovalent streptavidin–biotin interactions. Recombinant proteins can be site-specifically biotinylated to enable high-affinity binding to QDs conjugated to unimonovalent streptavidin. (d) HaloTag–chloroalkane association. Proteins expressed as a fusion to a HaloTag domain covalently bind to chloroalkane linkers conjugated to QDs.

Other types of recombinant proteins may afford a more stable association between a QD and its protein target. One of these approaches exploits the highly specific binding between the bacterial streptavidin protein and the small molecule vitamin biotin. The streptavidin–biotin interaction is the strongest non-covalent association known, and is frequently employed in a variety of molecular biology applications. Ting and co-workers have developed proteins that are fused to peptides that can be biotinylated under physiological conditions⁸. Thereby, QDs conjugated to streptavidin can bind with high affinity and specificity to these engineered proteins (Figure 3c). Streptavidin is natively a tetrameric protein with four biotin-binding pockets, but protein engineering has enabled the isolation of a monovalent streptavidin variant, and monovalent streptavidin conjugates of QDs have been prepared with precisely one biotin-binding site⁷. Another strategy is the use of HaloTag proteins, which are haloalkane dehalogenase bacterial proteins that have been mutated to readily form a covalent bond with chloroalkanes⁹. Because chloroalkanes are very rare functional groups in biology, one can label a HaloTag fusion protein with QDs that display chloroalkane groups (Figure 3d).

Receptor dynamics and tracking

The cellular plasma membrane is a complex fluidic surface rich in microdomains that serves as a substrate for receptor-mediated signalling events. Dynamic diffusion and trafficking of membrane receptors has been found to play a pivotal role in many cellular processes, but the detailed dynamic events remains poorly understood. This is because researchers have traditionally studied these receptors through bulk-averaged fluorescence from many dyes and fluorescent proteins that are bound to receptors. Single-molecule imaging of dyes bound to receptors has revealed that these processes and behaviours are heterogeneous, changing over time and space and from one receptor molecule to another, but these studies have been limited to fairly short durations (<5–10 s) because of dye photobleaching problems. Previous studies using QD–ligand or QD–antibody conjugates revealed the complex workings of single membrane receptors at high sensitivity and temporal resolution^10,11. New receptor behaviours have been reported such as motor-driven transport of the epidermal growth factor receptor along cellular outgrowths toward the cell body⁹.

Neurobiology is poised to greatly benefit from receptor tracking experiments with QDs. This is because neurons have richly complex plasma membranes with multiple types of microdomains that form intracellular signalling complexes called synapses which are known to exhibit dynamic receptor exchanges. The diffusion of several types of neurotransmitter receptors into and out of the synapse has recently been studied using QDs attached to glycine neurotransmitter and α-amino-3-hydroxy-5-methylisoxazole-4-propionic acid (AMPA) glutamate receptors, revealing rapid fluctuations in diffusion rates in different membrane regions^10,12 (Figure 4). Some receptors, such as the neuronal growth factor (NGF) receptor, become internalized into the cell once they bind to a specific ligand, a process that can now be studied in great detail due to the photostability of QDs¹³. QDs conjugated to NGF were found to bind to the NGF receptor in the terminals of neuronal axons (long neurite outgrowths involved in signal transduction), which induced endocytosis of the receptor–ligand pair within vesicular structures. QD imaging revealed that these vesicles usually contain only a single NGF receptor, and are shuttled great lengths to the cell body along multiple molecular tracks that behave like a multilane highway within the axons.

Figure 4.

Imaging of red QDs bound to glycine receptors on the plasma membrane of a neuron. Arrows indicate clusters of receptors on cellular outgrowths. The microtubules of the cell are stained green for contrast. Figure reproduced with permission from reference ¹⁰.

These initial studies were carried out by using conventional QDs with large sizes and multivalent ligand presentation. Researchers have found that the large size of conventional QD probes is a major problem for their applications in crowded cell-surface domains, such as the synaptic cleft, an intracellular junction between neurons that is typically only 20 nm wide. Larger QD conjugates have limited access to this region compared with smaller antibody–fluorophore conjugates, which adds a certain degree of uncertainty to the some of the neuronal diffusion studies undertaken to date using QDs^3,14.

Intracellular QD delivery

One of the most promising applications of QDs is for the study of protein–protein interactions and dynamics inside living cells. However, a major challenge is to develop innovative methods for loading or delivering freely diffusing and monodispersed QD probes into the cytoplasm or other organelles of living cells. So far, the most effective way to deliver any type of extracellular cargo is to inject the substance directly using a microneedle. However, this process is rather low-throughput since individual cells must be injected one at a time¹⁵. In order to achieve higher-throughput delivery of QDs to cell populations, other efforts have been made to temporarily permeabilize the cellular plasma membrane through the formation of microscopic pores, either through the use of bacterial toxins (e.g. streptolysin O) that form well-defined membrane pores or through brief exposure to a pulsed electric field (Figure 5). These mechanisms are promising, but have yet to demonstrate homogeneous delivery of free QDs in cells. An alternative and highly promising approach is the controlled disruption of endosomal vesicles. Cells naturally engulf their surrounding environment through a variety of processes that yield intracellular vesicles containing extracellular fluid. This is a convenient mechanism to enable entry of QD probes into cells, but the particles remain trapped and are therefore not free to interact with their target molecules, so it is necessary to have a strategy for QD release or endosomal escape. One method is to use osmosis for swelling and bursting the endosomes¹⁶. This can be performed by allowing cells to engulf QDs during a brief exposure to a hypertonic medium (prepared by adding sucrose or other solutes), which leads to the rapid formation of pinocytic vesicles that bud off of the plasma membrane due to water moving out of the cells (efflux). In a second step, a brief and well-controlled exposure of these cells to a hypotonic solution containing a low solute concentration will cause water to rush into the solute-rich vesicles, inducing osmotic lysis and allowing the QDs to be dispersed into the cytoplasm. Another mechanism is through chemical transfection^17,18. A variety of polymers and cationic lipids have been found to efficiently disrupt endosomes or cause vesicular leakage following endocytosis, allowing efficient delivery of macromolecules such as DNA and proteins. Although the mechanism controlling the release of cargo from the endosomes is not well understood, it is likely that the transfection agent also induces osmotic swelling of the intracellular organelles.

Figure 5.

Methods and mechanisms for intracellular delivery of QDs (red). QDs can be injected directly into the cytoplasm or nucleus, and QDs can naturally diffuse into cells after the formation of temporary pores induced by an electric field or pore-forming bacterial toxins. In addition, QDs can be loaded into cellular vesicles through endocytosis or osmotic pinocytosis, and then released into the cytoplasm by osmotic swelling and vesicle rupture.

Intracellular diffusion and targeting

QDs have been widely used for visualizing the dynamic motion of motor proteins. Inside cells, proteins such as myosin, kinesin and dynein transport cellular cargos to specific subcellular locations by moving along filament structures through a process powered by ATP hydrolysis. Motor protein translocation has been observed to proceed in discrete steps and with a velocity indicative of specific motor protein–filament pairs. QDs have become important tools for the study of isolated motor protein function outside cells using light microscopy, and they have recently been used to study these same phenomena in living cells. QD–kinesin and QD–myosin conjugates delivered to the cellular cytoplasm through osmotic pinosome lysis were found to undergo directed motion inside cells that was remarkably similar to that observed in purified filaments^16,19. Single-molecule imaging of these conjugates revealed an exceptional level of spatial resolution, and the motors could be tracked for extended periods of time without loss of signal. However, these intracellular experiments have been hindered by the large size and multivalent nature of conventional QD probes. This is primarily because the cellular cytoplasm is a crowded solution of macromolecules amid a network of filamentous structures that acts as a molecular sieve. Conventional QDs are larger than the sieving threshold for most cell types and exhibit dramatically reduced diffusion. This has not been a major problem for studying plasma membrane proteins because the viscosity of the plasma membrane is the limiting factor in diffusion, and cytoplasmic motor proteins demonstrate directed motion, not free diffusion.

Figure 6 shows an experiment in which we loaded cells with three sizes of nanoparticles, each with a distinguishable fluorescence colour, and we monitored their diffusion in different parts of the cells. Only the smallest QDs exhibited diffusion similar to globular proteins and native cellular components. QDs larger than 20 nm showed dramatically reduced mobility. A second major limitation of current QDs for intracellular imaging is the rapid fluorescence blinking of single QDs. That is, the fluorescence intensity of an isolated QD fluctuates between ‘on’ and ‘off’ states. This effect is averaged out in clusters of multiple QDs and in large ensembles, and is a useful means to identify isolated single molecules, but can disrupt the time-dependent observation of dynamic cellular events. This is not greatly deleterious for plasma membrane proteins, where the mobility is significantly reduced and diffusion is confined to the focal plane of the microscope, but this will significantly hinder QD tracking in three dimensions in the cytoplasm, where the loss of QD signal could either result from blinking or from diffusion of the QD out of the focal plane. Future success in QD tracking in intracellular locations will require QDs with reduced blinking; efforts to engineer QDs to eliminate blinking are currently underway²⁰.

Figure 6.

Effect of QD size on cytoplasmic diffusion. Three different sizes of fluorescent nanoparticles were loaded into cells through osmotic pinosome lysis, each size coded with a discrete fluorescence colour for identification. A bright-field (BF) image of the cells is shown on the top left with an overlap of a Hoechst dye that stains the nuclei blue, showing two adjacent cells. Three fluorescence images depict the three different colour channels for each nanoparticle size (40, 28 or 10 nm). The insets show magnified views of the trajectories of an individual QD in the same cellular region, marked by a yellow box. Only the 10 nm QDs show unrestricted Brownian motion; the 28 nm and 40 nm particles are highly confined in space. Scale bar, 10 µm. Data provided by Andrew M. Smith, Michael C. Mancini and Shuming Nie.

Future directions

Single-molecule studies have revealed that the microscopic world of biology is considerably more heterogeneous and diverse than we could ever have imagined. The ability to image and tracking single macromolecules is thus essential for a mechanistic understanding of cellular biology. Recently researchers have shown that the diffusion of more than 1000 QDs conjugated to plasma membrane receptors can be simultaneously imaged and charted on the cellular plasma membrane to generate a diffusional map of the plasma membrane, which is shown to consist of distinct domains with unique properties that fluctuate over time²¹. This cellular and molecular mapping approach could be applied to different compartments in cells to reconstruct the entire structural and functional framework of individual proteins inside living cells in real time. Projects such as these will be aided by the unique properties of QDs that not only enable simple single-molecule imaging, but also allow the simultaneous detection of a large number of unique spectral colours to code multiple proteins. These approaches could apply not only to individual cells isolated in cell culture conditions, but also to living cells in living organisms. Indeed, single-molecule tracking of plasma membrane receptors has been demonstrated with in vivo microscopic imaging for the study of tumours in animals²². Another exciting direction is the use of multicolour QD probes for super-resolution optical imaging of dynamic cellular events²³. Conventional optical microscopy is limited to a spatial resolution of around 200 nm (the diffraction limit), but researchers have recently overcome this limit using several approaches such as stimulated emission depletion (STED) and photoactivated localization microscopy (PALM). These techniques greatly benefit from the capacity to optically switch the fluorescence of a molecule from an ‘on’ to ‘off’ state and vice versa. An important step in this direction is to develop ‘switch-able’ QDs and strain-tuned type-II QDs (with minimal overlap in their fluorescence and absorption spectra). Indeed, the capacity to image the real-time dynamics and functions of single molecules with molecular resolution would open a new window into molecular and cell biology.

Biographies

Andrew M. Smith is a Distinguished Fellow of the NIH Center for Cancer Nanotechnology Excellence at Emory University. He received his BS degree in Chemistry and his PhD in Bioengineering, both from the Georgia Institute of Technology. His research focuses on nanomaterials engineering for molecular imaging of cancer and the exploration of the interactions between nanostructures and biological systems.

Mary Wen is a graduate student in the joint biomedical engineering program of Emory University and Georgia Institute of Technology. She received her BS degree in biomedical engineering at the Johns Hopkins University. Her research interests are in the development of new tagging strategies for size-minimized QDs and their applications in molecular, cellular and in vivo imaging.

Shuming Nie is Wallace H. Coulter Distinguished Chair Professor of Biomedical Engineering at Emory University and Georgia Institute of Technology. He received his BS degree from Nankai University, his MS and PhD degrees from Northwestern University, and did his postdoctoral training at both Georgia Tech and Stanford University. His research interests are primarily in the areas of biomolecular engineering and nanotechnology, with a focus on bioconjugated nanoparticles for molecular imaging, molecular profiling, pharmacogenomics and targeted therapy. snie@emory.edu